Veritone, Inc. (VERI) Earnings Call Transcript & Summary

Good morning. Welcome to Veritone's Investor Day and Tech Expo. I'm Brian Alger, SVP of Corporate Development and Investor Relations. We're excited to get started as we have a jam pack lineup for you all today. But before we get going, I need you to take a look at the legal disclosures. As you can see from the slide, we'll be making forward-looking statements today. These statements are subject to risks, uncertainties and assumptions that may cause the actual results to differ materially from those stated or implied by the statements. The risks and assumptions are detailed in our SEC filings, including our Form 10-K. The statements are also based on the assumptions as of today, May 14, 2021, and Veritone takes no obligation to revise or update them. All right, here we go. Today, we're going to introduce you to a number of our BU leads, their team members and a number of our partners and our customers. We plan on having 4 Q&A sessions over the course of the morning. Now I know that questions can be entered at any time in the questions box on the left of your screen. I'll be moderating the Q&A sessions, and I'll be communicating each of the questions. The first Q&A session is going to be after GLC. The second will be after our aiWARE Labs discussion, which will be followed by a brief intermission. Then we'll have another Q&A session after our Media offerings, and then a final general Q&A segment when Mike wraps up with the financial comments. All right. With the logistics out of the way, let's get started. It's my pleasure to introduce our Chairman, CEO and Co-Founder, Chad Steelberg. Chad?

Thanks, Brian, and thank you, everyone, for being here today. Welcome back to those of you who have been here before. I would like to take a brief moment to acknowledge the braveness of the human race. Our frontline workers and the perseverance of technology innovators to help us solve today's global challenges, and move beyond the pandemic back to normal life with our families, friends and colleagues. One year after closing our offices because of the pandemic, I'm thrilled at the strength of our internal culture and the productivity of our remote workers. We are focused on our VERI values to improve our internal culture, management engagement, local communities as well as our new talent recruitment. Today, we are going to focus on both the technology and market trends that have enabled several quarters of tremendous growth for Veritone and what looks like continued hyper-growth for the rest of 2021 and beyond. This slide lays out the agenda for today. As you can see, we will hear from each of our business units, have some lively Q&A, introduce an exciting new product and then Mike is going to outline our long-term financial goals. Staying true to our vision to democratize artificial intelligence for the world through the first artificial intelligence operating system, we are seeing phenomenal product momentum in the market and are now adding professional services and system integrators to expedite global adoption and large enterprise engagements. We are thrilled with the adoption of our core solutions like aiWARE and Studio, along with our industry-leading applications and solutions. Our customers are benefiting from the underlying power of aiWARE, as they expand the use of AI in their enterprises from an initial entry application to now broader solutions and customizations, where they are seeing the value, flexibility and scale it provides. Equally important, system integrators are seeing the value of aiWARE to enable faster business transformation and long-term value for their customers. Teaching and enabling companies to build their businesses using aiWARE as a foundational tool is a key focus of the company, and we are excited to see these partners starting to succeed. [Presentation]

And that is just a fraction of what's in-store for you today. I cannot tell you how excited I am about the adoption of both customers and partners and the advancement of our technology portfolio. Today, we will walk you through our business units, including government, legal and compliance, aiWARE, Veritone Labs and Energy as well as Media, which will include a revolutionary product launch. Investors and customers often ask me what will Veritone look like in 5, 10 or even 25 years from now? The answer to this question provides profound insights into Veritone. Our understanding of artificial intelligence and my goals and aspirations for humanity and in the end, the value we are creating for our shareholders. First, I think we all have to agree on the importance, longevity and impact of AI technology. If you've been paying attention, you've most likely heard nearly every technology executive and futurist, including myself, declare AI as the most powerful and impactful technology in the history of mankind. But why is that? My perspective on this is foundational to understanding Veritone as it shapes our strategy, the products we build, who we hire and how we operate on a daily basis. So what makes AI so damn important? It is really just 1 simple thing. AI is the only technology that fundamentally changes the rate of human innovation. Human innovation is defined as the rate at which we alter our society through new ideas, products, services and even our actions. It has been studied by scientists for years, and we now understand that the rate of human innovation is governed by 2 primary factors: the first and most important factor is simply the number of humans on the planet at any given point in time. And the second is the speed by which they communicate with each other. So improvements to communications. Things like speech, the written word, the printing press, the television, telephone and the Internet are dramatically increasing not only the speed, but also the reach every human has to communicate and share ideas with nearly every other human on the planet. People often say that it feels like the world is moving faster and faster, and it is primarily because we've increased the speed and reach of communication by several million times over the last century. But in the end, despite advancements in the rate of communication, it's still the number of humans on the planet that's the limiting factor until now. With AI, for the first time, we are adding a new cognitive dimension to the equation that governs the rate of human innovation, where ideas can now be generated independent of humans and can be shared nearly instantaneously between these new cognitive AI machines. So as AI technology improves and the number of AI machines grow, we are going to see the rate of human innovation explode; pulling forward innovations that were hundreds of years in the future to within our lifetime; medical miracle cures to space exploration that were unimaginable are now just around the corner. Do you really think that mankind is going to explore interstellar space? Do you really think that the podcast you'll be listening to in a few years will be produced by a real person? Will your grandchildren drive a car? Can we stop climate change and mass pollution in time? Can we reverse it? What about social unrest, human bias and systemic racism? Artificial intelligence holds the key to all of these and so much more. So with that foundational belief in AI, the question of where does Veritone fit in becomes even more interesting. The answer to this question begins with a thought experiment. It goes something like this. I want you to think of the first company that comes to mind when I name an industry. For example, if I was to say, online search, you would almost likely be thinking of Google. So let's give it a shot. Online shopping? Amazon. Good. I see you're all getting the hang of this. All right. What about online car service? Uber, maybe some of you said Lyft. Online auction? eBay. Okay. Here's a good one. Online social, video? YouTube or maybe some of you even said TikTok. We could keep playing this game for hours. As you can see, each industry quickly standardizes on just a few companies that become synonymous with the category. The network effect and the relative scale of these leaders creates a force that once started becomes nearly impossible to stop or slow. So in 5, 10 or even 25 years from now, when someone says the phrase, artificial intelligence, I can assure you that only 1 or 2 company names will come to mind, and I want Veritone to be the first. Given the profound implications and power of AI, the only way to achieve this audacious goal is to build a cognitive operating system, which allows AI to move freely from one industry to the next, from one application to another, supporting every AI model on a common data backbone, and independent of any cloud provider or hardware infrastructure. Truly, one AI OS to rule them all. We are clearly not playing for second place. We are not building bespoke one-off solutions on open source software that won't scale or keep up with the latest innovations. We have built a foundation in our operating system that continuously improves and advances applications without disruption. We are democratizing the development and access to the best AI engines and models at scale. We are changing the rate of human innovation. I hope you enjoy the next few hours and find yourself inspired by the progress we've made towards achieving our vision of building a more just, innovative, transparent and vibrant society through the power of aiWARE. With that, I would like to hand the call over to Jon Gacek, who will introduce government, legal and compliance. And I will join you again soon during the Q&A session. Thanks again for your time and interest in Veritone.

Welcome to the government, legal and compliance section of Tech and Investor Day. My name is Jon Gacek. I manage and lead the GLC team focused on this market. We've got about a 25-minute presentation today, mostly full of customer testimonials and tech demos. But I want to set the stage for you on how we look at the markets, how we go-to-market and what we're selling. We think of GLC as 4 separate markets. The first is legal, and here, we're focused on improving the overall experience for legal review, making it more efficient and more cost-effective by cognitively enabling a lot of the processing. And what I mean by processing is looking at evidence and using AI to do so. Our customers are MSPs, managed service providers, who tend to be providing services to legal customers, both law firms and corporate end users. The second grouping is SLED, state local government. We are very focused on law enforcement, either at the local police agency, county police or state police, and then district attorney, public defender, across the counties and cities, and then we get up into states across state agencies. Fed is a little bit different. We think about Fed in 2 separate groupings: civil and DoD intel. For civil, we're focused on the DOJ, DHS are 2 of the largest, but also EPA, IRS. This is the nonmilitary grouping. And what's interesting about this particular space is that it looks very similar to legal and SLED, in that we are selling and positioning our solutions for the same use cases, and we get a lot of leverage as a result. And I'll show you more about that in a second. For the DoD intel side, this is a little bit different. This is really leveraging our capabilities to be very portable, deploy aiWARE in multiple scenarios, whether it's behind a firewall in a cloud-native environment for one of the government organizations. We've actually put some proposals in that we're going to hear about where we're taking aiWARE and taking it down to the soldier, to a soldier as a sensor where you can carry our technology as an individual. And here, we position differently, but the use cases and the technology that we're bringing to bear are all based upon the same foundation, which is aiWARE. And aiWARE is common across all 4 of these markets. And it's a place that allows us to get a lot of leverage. And so what we've done in the GLC group is we've built a series of applications on top of aiWARE using our own tools to solve very specific problems for this set of markets. And as you look at this slide, you'll notice that there's commonality. Illuminate shows up in all 4 markets. It gives us -- it is a very powerful program, and it gives us a lot of flexibility. Redact shows up across 3 of the 4. And so one of the things that I want to have as a key takeaway from this session is, this technology that we're selling is based on aiWARE, and we get a lot of leverage across all of our markets. Another important leg on our stool, though, that I want to spend a minute on is go-to-market. And we think about go-to-market in these 4 segments. First, our cloud partners. We have a very close relationship with Microsoft and AWS. We are FedRAMP authorized technology partners. It's interesting -- this was a little bit unexpected. As we came out with applications, we found that we had very good synergies for solving problems with a group of technology partners. These are where -- these are the ones we have partnerships with today. One of the ways you'll be able to measure our progress in advance of revenue will be announcements around new partnerships. We will have more of these. We'll continue to expand them, but it's become a critical way for us to get to customers and provide them a unique solution. The third grouping is managed service providers and system integrators, critical in the legal space, but even more critical in the Fed space. We are partnered with some major firms who use our solution to create very, very unique value for their customers. And then the final group is resellers and distributors. We've built some great partnerships with CDW, with Carahsoft, with WSI. These folks are reselling our product, carrying it on their price list, and it's important in this space because they often have access to funding vehicles and some of the contracts that are required to grow this overall business. So from here, I'd like to jump right into product demos. Ben Ha is going to take over from here and do a demonstration of our Illuminate product. Thanks very much.

Hi. My name is Ben, and I'd like to introduce you to Veritone Illuminate. This application is purpose-built to rapidly process a large corpus of data, which often requires a lot of human review time. For example, if you have thousands of hours of audio or video you have to go through or millions of images collected off a hard drive, you can use Illuminate to quickly search, call down and take action on that responsive data. So let's get into how Illuminate works. At the very top of the screen, you'll see the search bar, which allows us to now search through all the content that has been brought into the platform and processed through AI cognition. This first example is searching by keywords. So if I have audio-based content or video containing audio, I cannot quickly search through all that data, maybe hours and hours of jail calls, for example. So I'm going to search for the phrase, kill them. And in this example, I'm looking for that confession across all that audio based content. You'll see this interview room recording is returned. And if I pull this up, I get that side-by-side playback view of the interview itself and the time-synced transcript. [Presentation]

So that karaoke style playback makes it really easy to follow along based off the audio content. And what's nice about this entire transcript is it is time correlated. So if I have another part of interest I'd like to click into, I can just jump right to that word and start playing it back from there. Now let's say that you're dealing with foreign audio, so beyond just transcribing in English, let's say, I had Spanish audio, like we're seeing here. What I can do is run transcription, like we're seeing here, but then what I can do is also translate that results in transcript back to English. And so now I have the translation side-by-side with the original transcript in Spanish. And I can search on that resulting translation to make it really easy to find my responsive content as well. So we've talked a lot about how we work with audio based content. On the visual side, what we can also do is search for other things on screen through, let's say, video and images. So let's say that you have a person of interest you're trying to find across hours and hours of films that you've collected. What I can do is simply type in the name of that person. And what I've done is actually train our facial recognition engine to find this person based off of 1 or more photos. So now when I run this search, it's going to go through all the video and all the image in the system, and now we're able to quickly narrow down the resulting files where that face was detected. And so now you can see I can jump right to the 92nd mark on this video. And I can start playing that back just like it could on the audio side. So again, everything is time correlated, making it really easy to review that content. Now once you've narrowed your content down to the relevant subset of responsive data, you can then select those files and take action on them. For example, I may want to bulk tag these items so I can search and pull them up later on. And you want to bulk export this content out, so maybe my end goal is to provide this information to the D.A. or someone else or a further legal review within an eDiscovery platform, for example. And so now you can bulk export out the original files, the related assets, like the transcripts, and you can hand those over. A third option is sending into our Redact application. So with Veritone Redact, which we'll show you later today, we have the ability to quickly redact the audio and video-based content and be able to turn that over as part of the full year request. That concludes the overview on Illuminate.

Ben, thanks for the demo of Illuminate. Let me give a little bit of context. As Ben mentioned, this is a very broad solution. It is a great solution to position inside of accounts. And it has a lot of utility to it. And what you'll find in our next speaker, we started a partnership with WSI around actually interview room recordings. But when they get into accounts and start talking about our joint solution, almost invariably, they start talking about the other Veritone applications. And one of the things our customers really like is our ability to switch between our applications, all powered on top of aiWARE.

Hi. I'm Joe Fitzgerald, a sales engineer with Veritone's government team. I'd like to introduce you to Mitch Thompson, Director of Public Safety Solutions at Word Systems.

Before I came to Word systems, I was fortunate to serve in public safety for 28 years, retiring in 2018 as a Chief for Police. Today's agency is really trying to focus on being more transparent than what they've been in the past. And part of the transparency is having a lot of audio and video data in front of them and also other relational data. What they have found is they don't exactly know what to do with the data once they have it. And so the problems that they're faced with is there's this massive amount of data. They're trying to be better at solving crime with that data. They're trying to be more transparent with the community with that data. So they need solutions that help them get there. Look, law enforcement, historically, has always relied on human intervention to make things happen. And they're starting to realize that there is machine learning and artificial intelligence that can help them achieve the same goal in a much more rapid fashion. Where Veritone can bring to bear to law enforcement is the seamless transition of being able to identify evidence, being able to produce that evidence and then being able to publish that out to the community. And let me give you an example of that. So using the Veritone features, you can use that for facial recognition to recognize the suspect of a crime and you obtain that video to do that. From that, you bring in that suspect and then you're conducting the interview. And from the interview, you uncover not just that crime, but perhaps multiple crimes. And with that, you're able to tag and mark and search all within the transcription and go to those exact moments in the video file and perhaps make an arrest for prosecution. And then the community wants to know about this crime. And so you have the ability then to go in and provide a video that is releasable to the community because you can't put all that out. And the ability or the power that Veritone has is to make that a seamless transition across the whole -- entire process so that it's not jumping from one engine to the other. We know behind the curtain it is, but the client to them is just input in and then they get the output.

Thanks, Mitch. We really appreciate your comments and taking the time. And our partnership with WSI and your team is fantastic. One of the things I want to emphasize for the listeners is what we're talking about here is we go in with a very pointed solution. We get a meeting, but invariably, these customers want to talk about the rest of our solutions. And that's one of the things that we're very excited about. And one of those solutions is our Redact application, which Ben Ha is going to demonstrate for you now.

With the proliferation of body worn cameras, agencies are subject to more and more public records request with this short turnaround time. And these times are often mandated by state legislature. This places an undue burden on the records management team, and that's where Veritone Redact comes in. Using the latest in AI technology, it helps automate the video and audio redaction process while still allowing the user to supervise that final redacted output before it's turned over to the requester. So without further ado, let's get into how Redact works. What Redact will do is present the review results of the automatic AI detections that we talked about earlier. So here, we see that heads of people are shown as well as laptop screens, license plates. You can even create custom user defined regions, in yellow for virtually any other object that has to be obscured. And so now as a user, you can quickly jump to those different moments in the video and check or uncheck whether or not you wish to apply or give a redaction. Let's take a look at how this actually works. What I'm pulling up now is an example of a video that's already been processed through that automatic detection. And I'm going to playback a short clip of this, so you guys can see how this works. So you see those green detection boxes popping up on screen for those heads of people. And you'll see I can click right into any one of these, simply check or uncheck whether or not I wish to apply that redaction based on my local redaction requirements. What you'll notice here is that we're actually doing what's called head detection, where we look for a head on a torso and where that provides a lot of benefit is in situations like this where the face is too far away or blurry, or here where the persons back is to the camera or nowadays people have face masks on. In all the situations, Redact can pick up those without any sort of issue, and it helps you to be a lot more accurate in terms of those automatic detections. Next, we're going to take a look at how audio redaction works. So now we have the ability to go in and transcribe that spoken conversation and use that transcript as a visual map to quickly find and remove sensitive audio. So one of the options is searching or we can even skin through that transcript. [Presentation]

And then click on a word like I just did, and then we'll start playing back that audio. And now we're able to use that audio to quickly find that sensitive information. So once we validated that the information is sensitive and has to be removed, I can simply highlight that part of the transcript, right click and redact it. And now it's going to automatically find that start and end time, in this case, a 6 till the 9 second mark, and it's going to mute or turn over that audio for me. So it's a visual approach to an audio redaction. I can also search for a word or phrase, find every instance of that throughout the transcript, right click and redact that. And so now if I play this back, you'll hear how this sounds. So it's muting over that sensitive audio and then it starts playing back again. So what we found is that this approach to video and audio redaction is a huge timesaver. If you look at the traditional frame by frame video redaction approach, an hour long video takes anywhere from 5 to 10 hours to complete. With Redact, we're able to cut that time down to under an hour on average. So it's closer to a 1:1 based off the length of the media file. Once we're ready to apply those redactions, we can click the redact file button. And what that's going to do is generate that published redacted output for review in this tab. [Presentation]

So you can see how it's following along and blurring that head. That's actually a level 10 blur, which can be adjusted to be lighter, if preferred. And then for the audio, you heard how it muted over that sensitive portion, you can also change that to a beep or toning noise to make it more explicit. When you're ready, you can click the download button. And what that's going to do is produce not only the redacted audio or video file, but it includes a chain of cuts to the autolog. So that's going to capture the date and time of the action, the user that was logged in and the details of every action they took within Redact.

Ben, thanks very much for the demo. Hopefully, you in the audience can see the power of that solution, how much time it can save and how efficient it can be by using AI technology running on aiWARE. We have great product market fit with Redact, and it's one of those products that customers find very, very valuable.

Hello. My name is Aviv Redlich. I'm on the inside sales team here at Veritone overseeing the West Coast. I'm excited to introduce you to Amanda O'Neill. Amanda is the record supervisor at the Eureka Police Department. What you're going to hear is Amanda's experiences using a legacy audio/video redaction application, and how Veritone was able to help her expedite that process of redacting audio and video.

We have an evidence technician who strictly focuses on digital media. And she would utilize the software like a platform within the body 1 system she'd go through and redact footage. And it was a frame-by-frame redaction. So she would physically have to select 1 screen, 1 frame, edit, move on to the next. And I think her worst -- her biggest recording up to the time we found Veritone was 10 minutes of recording, and it took her full 8-hour workday. How are we going to prepare videos screen like frame by frame? And is there a company like do I just shift this to someone? And then these are like sacred moments in people's lives. You can't just turn this over to an anonymous person to say, "Hey, please redact this for me." So just the marriage of technology and AI with us having that in-house control is huge. It was actually a webinar that I was first introduced to Veritone on. I was listening to it, and I'm just thinking, "Oh, my gosh, this is something I just never would have put in the realm of public safety." And yet as soon as I started hearing more about the company and more about how AI is used, it just made complete sense. I remember getting off of the demo and walking into our chief's office and saying, "This is it. This is the coolest stuff like I didn't know that we could utilize AI." I didn't -- like, "They've done the work for us. We just have to basically upload videos, hit a -- click a couple of buttons, and we're done. We're good." The time that saved is incalculable for me. I can -- for most videos, I can truly click a few buttons, maybe go in and polish just a brief shot or 2, and my files are finished, which is amazing. Like it seems like it should be something like in a Sci-Fi movie, but I'm actually able to use it here in Eureka. So it feels like we've always had this resource and yet it's really been a resource that I had for a year, and I can't fathom not having. So like -- yes, thank you.

Amanda, I'd like to thank you. I really appreciate the time you spent with us and the passion that you show for what we're doing. And we look forward to supporting you in the future. Redact is a very strong product. We have great product market fit. And we have lots of partnerships where we are combining Redact with other people solutions. One such company is GovQA. GovQA has an excellent FOIA software. It's based in the cloud. We have very good alignment at the senior team level. We go-to-market together in the field, and we provide a very unique solution to our customers and our prospective customers.

Hello, everyone. My name is Kyle Dayson, and I'm an account executive on the GLC team here at Veritone. I'd like to introduce you to Phil Hodge, who is the Chief Strategic Development Officer at GovQA.

There's 3 things that I'd point to when it comes to working with Veritone. One, Veritone is a team of technologists. They understand and believe in the power of technology to solve real problems and their Redact problem -- or the Redact product and the problem that Veritone and GovQA solve for our mutual customers is wonderful evidence of that. They are an organization that loves to be customer-facing and work the deal, which we share that sentiment, and it creates a wonderful partnership between the 2 organizations. And the final thing I'd point to is Veritone is an industry leader, right? And as the industry leader for public records, requests and automating that, it's just been a pleasure for us to go to market with another industry leader. Video review and redaction today is a time-consuming process. Again, when we look at our peers in public records index, we see the obvious trends. Most notably, the volume of video files is increasing exponentially. Using a manual process, like most of our agencies use today, that correlates to a direct increase in the amount of time and resources required to process a request. At the same time, the laws that require timely responses are not changing, and that creates significant challenges for all of our customers. Thankfully, Veritone and their democratization of AI offset this challenge by making the process more efficient, allowing our users to handle these videos in a shorter amount of time, yet maintaining accurate and comprehensive delivery of relevant records in a compliant manner.

I want to thank Phil very much for taking the time. He and I have a weekly call to go through deals and opportunities. And so we spend a lot of time together. And I appreciate his comments and how his firm looks at the partnership between Veritone and GovQA. Now I want to transition to another partnership type, which is an SI. Deloitte is one of our longest and largest partners. We are very integrated into their legal team. We are jointly going after business together and solving problems for the DOJ, and there's a lot of opportunity and a lot of activity between us and Deloitte.

Hi. My name is Karenna Soto. I'm an account executive at Veritone. I'd like to introduce you to Pat McColloch, Managing Director, government and public sector, Discovery at Deloitte.

When Veritone and I first started meeting several years ago, I stressed to them the growing concerns of data security in the government space, and how the government was really focusing on FedRAMP-approved solutions. FedRAMP is the government's program that they use to audit and certify solutions that are going to store government data outside of a government environment. Veritone really took those messages to heart and made the investments to make sure that their products could be used in multiple environments, including a service provider FedRAMP environment, such as Deloitte's, or inside one of our government client systems, or they also built their own FedRAMP-approved cloud environment, which was the first operating system for artificial intelligence to meet the FedRAMP designation. Our clients' data is growing more complex in size, scope and complexity every day. Gone are the days of litigators, investigators or analysts being able to read every document, listen to every recording or watch every video. The application of AI and machine learning is an absolute table stakes tool that has to be brought to solve these challenges. Today, Veritone is providing those capabilities to Deloitte and to our federal clients.

Pat, thank you very much. As always, we appreciate the time and the kind words. One of the things that's unique about Deloitte is that we have the strongest partnership with them on the civ side of the federal government, but we've been active with them on the DoD and intel side as well. And SIs are very, very important in both civ and DoD, and it's great to have a partner like Deloitte, who's interested in creating solutions with aiWARE and Veritone on both sides of the Fed business. What I'd like to move to now is focus specifically on DoD and Intel. And I'm pleased to introduce Lisa Marie Cheney, who is part of our Advisory Board, who has spent -- spends a lot of time with me talking about opportunities and strategies for penetrating the DoD and intel space. Lisa Marie has had a long and illustrious career in supporting various administrations, specifically on the DoD side. And without further ado, I'd like to introduce Lisa Marie Cheney.

When I served at the Department of Defense as the Principal Deputy Assistant Secretary Defense for OSD(LA), the big movement there was asymmetric warfare. We were moving to a surge in Iraq and Afghanistan. Today, I think if you bring us forward almost 15-plus years, the big movement is towards AI and ML. Veritone is poised at that cutting-edge right now to be the leader and the game-changer and the thought leader for that. And if you look over our history, we went from mobility by foot to horse to car, to rocket and now we're looking at aiWARE, which is going to allow us to go from just simple processing, using our computer where we thought, "Hey, we got windows, and we haven't changed the world." Now if we can take all of the data that we have and everything that we've learned and make -- put it all together, model out multiple decision factors and pick the best one that fits the policy of that decision maker, it's incredible. I mean, that is the next generation of where the world is moving. And again, Veritone is poised at the center of that to be that game-changer.

First, I'd like to thank the presenters. We really appreciate the time that you put into this, and I love the candid comments that you provided. I think it's important for our shareholders and our partners to hear from our customers. And again, I want to thank you all. And I want to thank Ben for the demos. As always, you did a great job. I had several points I want to emphasize and kind of key takeaways that I hope came out in the presentation, and I'm happy to answer questions at the end. The first is Government, Legal & Compliance is dominated by human processes and bespoke applications. We think we're in a unique position to create efficiencies, create new workflows, cognitively enable existing workflows and create a lot of value for our end-user customers. It's a very target-rich environment for us. And we find that as we get in front of customers, we have unique solutions that we can provide them. Number two, it's about partners. This space, you need partners, you need solutions. And if you can put together something like what we've done with GovQA or March Networks, you're going to drive more access and better solutions, and you're going to see us do more of this. And so I really want to emphasize the work that we've done here, and you'll see more and more investment from us as it relates to recruiting and partnering. Number three is really about deployment. Each of these market segments has unique deployment requirements. In legal, it's all about security, in SLED, it's about CJs. The fact that we've deployed in GovCloud both AWS and Microsoft is pretty unique. And on the federal government side, I want to emphasize how important our FedRAMP authorization is. We get incredible credibility, incredible credibility, I'm going to say it twice, because we're FEDRAMP authorized. And you'll see us add more sponsors here. It shows that we are very serious about our technology and security and it's a significant barrier to entry to anyone else. The third is really -- or the fourth that I want to mention is strategy. Our strategy is right. The opportunity is huge. aiWARE is differentiated. And our ability to move with this market to create cognitive solutions for our customers as we move forward here is -- it's a massive opportunity. I just can't emphasize it enough. And as we get into these customers and they see what we can do, they start thinking about what else can we do? I call it the Harry Potter moment. It is very, very important, but I want to leave you with, our strategy is right, aiWARE as a solution. We have technology now. Many of our competitors will sell a solution, but they're still creating it. They put butts in seats and they charge consulting hours to create things. We don't do that. We've got technology now, and we can go after solutions right now. And then finally, it's early. This market is very early. I really like how we're positioned. I like how quickly we are getting credibility in these markets. I like how quickly my team has been able to build solutions on top of aiWARE, and we can get them into market, and you're going to see us do a lot more. So I appreciate your time. We'll be around to answer any questions that you have. Look forward to talking to you soon. Thanks very much.

That's great. Great start to it, John. Good presentation. So this is our first Q&A session of the morning. We're going to compile those. You all should see a Q&A section on the left side of your screen. You can enter in your own questions into that, and then I'm going to moderate the questions as we go through. Remember, there are going to be 4 of these sessions throughout the day. So we're going to try and keep it to the topics at hand. In this case, we're going to kick one-off with a general question for you, Jon. Stephen Volz is asking a question that pertains to our task orders growing in size, specifically within the DOJ. And how some of those that are approaching 8 figures are coming to market and when we might see those materialize?

I'd like to see a lot of 8 figures. That's a lot of zeros, actually. What we're doing in the government space is very, very focused on building a platform and a business. And DOJ is the first large sponsor of us. I don't want to foreshadow too much. But because it's government, you'll probably hear about it before you see it on our income statement. It's a huge organization. You'll see a lot of activity, you'll see task orders issued. One of our key achievements this last year was getting on the MEGA 5 contract with some of the key partners there. There's 5 organizations that are part of MEGA 5. We are aligned with 4 of them. That's a $1.6 billion award that goes across those. We are starting to see task orders come out of that now. They won't start with 10 zeros, but they'll start with a lot of zeros. And we'll start actively pursuing those. So it's a journey. We're also focused on the other organizations as well. And everything moves a little slower in the government space. But as it starts to hit, it's a lot larger than in the commercial space.

Seeing things come in to the Q&A line here on the side, maybe while we're compiling those, Jon, can you maybe flush out one of the comments you made in your slide deck with regards to the leverage that you're getting across the 4 different segments of GLC. And why that leverage is enabled by aiWARE?

Sure. You saw in the graphic, aiWARE is the center of really what we're doing in GLC. And candidly, as you listen carefully to Chad's comments, it's the center of what we're doing at Veritone. And what we've done in a pretty small group inside the company in GLC is we used the capabilities inside of aiWARE to then build applications to connect to the various segments. And what we've picked our go-to-market solutions or solutions that have sort of a ubiquity across them. You saw in Illuminate what we could do there. Think about how powerful that is, you could search for a transcript for jailhouse recall, for an interview room, you could take video from a body cam, you take video from a CCTV camera, you could put real-time notifications on that, all of that is based upon the core technology of aiWARE. And I think it's very early in this space. And our ability to move quicker than everybody else and leverage technology again, aiWARE, is just an advantage in the government space. And so we're trying not to pick one offs. We're trying to pick solutions that are ubiquitous and can be used across. And Chad and I talk about this a lot. We don't really want to be an application company. We want to enable that. But we also want to connect to customers. And in the early markets, you need to have some applications to do that. I don't know Chad, if you want to add anything to that?

Yes, Jon, I mean, I think the one thing that I'm really liking right now that I see the market is customers -- we're meeting customers where they have their pain point. Some police agencies and law enforcement groups, their pain point is redaction and FOIA requests. Others, it's about evidence analysis to identify and solve nearly impossible cases, given the load of evidence. And so we can show up with an operating system and then bespoke applications that can meet them where they are. But very, very quickly because of the nature of aiWARE and the cloud infrastructure, those applications are now interoperable. So as you move from evidence discovery and prosecution into redaction, you're not switching platforms. You don't have to reprocess data. The efficiency that our partners are getting from that is really leading the charge in terms of how we're growing with our customers. And as we roll out new applications, either developed by our own internal teams or with our partners, like Deloitte, it really starts to create a marketplace and an ecosystem between not only individual companies and the applications, but between agencies themselves. And I think that's just a highlight that's really been accelerating over the past 12 to 18 months. And I think it's only going to grow.

And just to be transparent, you hear chad and I talk about this, but Mitch talked about it as well. And I thought Mitch was really on point. He's just out of the industry and once we get in, we can start talking about other things that we can do.

Great. Guys, we've got a question from Robert Johnson, asking for Jon to describe the competition and the budgets that you're competing against as they're looking to either establish new business or renew business?

Yes. It really depends. On the legal side, the competition will be humans or other bespoke solutions. We've got a nice cadence with our partners there. They understand what we do. They know we can move quickly. And so those are going to generally be one-off solutions that might have a transcription solution or might have a translation solution. In the SLED space, for Redact, in particular, it generally comes from the body cam companies. One of the value propositions for Veritone is we're completely open. So many of the body cam companies will only be able to Redact their own video. We can redact everyone's and across. So one of the big movements right now in public safety is drones. Drones create a lot of video and a lot of problems for the agency. That's a different video format than their body cam, we can bridge across those. So again, I'm selling a little bit here, but when you think about the competition, it's generally the body cam guys. And then I said it once, I'm going to say it again, on the Fed side, especially DoD, we're competing with butts and seats companies. One of our challenges is to go in there and say, "Hey, we've got a technology that's done. And we don't need to spend as much on butts and seats, which are expensive." And we're starting to see that turnover today. So some of those groups that are our partners, if they're not partnered with Veritone, they could be a competitor. And then on the civ side, it really looks a lot like legal. It's one-off bespoke solutions. I think we're getting a lot of traction in the DOJ right now because our ability to move across the entire DOJ and not have to have bespoke solutions.

Awesome. I've got a couple of questions from Mike Latimore. I'm just going to combine them here, Mike, hopefully, you don't mind. The first, Jon, is which of the 3 categories, legal, SLED or Fed, offers the biggest opportunities, specifically this year? And then as a follow-on to that, you talked about the deployment flexibility. Can you talk about how that's changed and how long it takes to get deployed today versus what it was just a couple of years ago?

Sure. I'm going to start with the latter, and then I'll go to the way and Chad, you can jump in here, too, because I know it's an important distinction for us. We started an AWS in the commercial space. And as we moved into AWS Gov, it was clear that we were going to have to do some re-architecting to make that successful. Then our next step was to move into Microsoft. And that's really where Chad and Alan team said, let's rethink about how we want to architect the solution and what we're going to do. And it took us about a year, candidly, to move on to AWS -- or I'm sorry, on to Azure. But after we did that, it became really easy to deploy the software across platforms. So we can stand up -- a customer can stand up on aiWARE edge today in a few hours. The deployment that we're talking about as a soldier as a sensor, that's a few hours. So we've moved dramatically. Our next big move here is to take the entire aiWARE stack, and I'm going to let Chad talk about this, and allow it to be deployed anywhere. And take the full functionality all the way down to your laptop or in something as big as a DoD cloud that's completely secure and let them run aiWARE behind the scenes. Chad, do you want to run with that? And then I'll come back.

Yes, absolutely. We've been pushing the aiWARE Anywhere mantra from a technology development road map. And at current, I'm proud to say that we are 100% cloud agnostic, but more importantly, we're completely infrastructure agnostic, which means I can literally today, in a matter of minutes, install a full stack aiWARE on my laptop. And now we're actually pushing into even smaller form factors like IoT devices for our edge processing. So we are pursuing a continuous infrastructure for AI analysis across all devices and systems, both Fed secure, all the way down to edge IoT devices. And this literally creates a backbone for the anybody building AI applications where they need cognitive services, can plug into that network in a standardized way regardless of infrastructure. We've achieved this in the last 18 months of effort, hats off to the team and the customers that are benefiting from that, Jon, mainly in your sector, it's great to see the adoption and the speed.

Thanks, Chad. I think in terms of Mike's first question, it was which market is the biggest opportunity in both short-term and long term? That's hard to say. I mean we have a lot of -- I use the metaphor with Chad. We planted a lot of fields. And you can't just drive by and see the fields that we planted. You have to actually fly over now, our planting has gotten that large. We have some really interesting contracts that we bid on. And the big scale is going to be in the federal government, for sure. I mean, those are the big, big dollars. In more of the masses, it's across SLED and legal. Our partnership with Microsoft, I often say, we're pretty happy when we add a new police agency, they take us into the entire state. And those deals just take longer, to get a whole state wide, every county in a state up on your solution, that's -- you're not going to get that done in 30 days. On the SLED side, we've got some real good product market fit. We actually closed deals in 24, 48 hours. And so think about as that's our recurring business. But our big, big 8-figure deals are going to be federal government first and then state government after that.

Okay. Great. Guys, we're going to try and keep things moving along here. So I'm going to combine 2 questions, 1 from Arthur Weise and another from Darren Aftahi, they're similar, asking about are we working with the body camera companies like Axon? And then if so, what percent of the police departments are using AI redaction versus -- and this is more to Darren's point, how many are still doing it manually? How big of a necessity is it to deliver the types of results that Amanda is realizing as they move forward?

Yes. So we can work with all the body cam videos. That, we can adjust in our system and redact it. Without -- I mean there's a lot of partnership things we could do with Axon. But today, at the Redact level, they tend to be a competitor of ours. They include their redaction solution as part of their overall solution. Most of the body cam vendors do that. Body cams still aren't proliferated across the world. And body cam has a lot of upside growth. I can't remember, the last status is like 40% and even a smaller subset are dealing with redaction. A lot of the redaction requirements are being driven by state laws. And so for instance, in Colorado, last year, they put out a new state law for every police officer had to have a body cam. And you had to have your video out within a certain number of hours. For 6, 7 months, we did no deals. For the last 3, we've done several multiple deals. So when I talk about things are just starting, the market is really ours to go after here. I mentioned drone cams. Some of the more progressive agencies are adding drones, completely blows up their model in terms of how they're going to redact that. There are some police agencies that have 30, 40 people who are doing redaction. Some of them spend $25 million on those people. If you think about what we're trying to do at Veritone, we implement an AI solution there. Now you can take those dollars and those people and focus it on the primary mission of a police agency. And again, that wave is just starting. The other thing that's kind of implicit in those questions is, people are going to chase us. We know that. But because of, again, aiWARE, our ability to do Redact 2.0 or Redact 3.0 to keep pushing the envelope, we can move much quicker than everybody else. And that's going to be a strategic advantage for us going forward.

Awesome. All right. Well, let's move forward. Our next segment is going to be presented initially by Al Brown. He's going to be talking a little bit about our technology, aiWARE, the Labs Group and the energy section is in here as well. Al?

I'm Al Brown. I am Head of Engineering for aiWARE. Thank you for taking the time to go over aiWARE and let me share this with you. Unstructured content is growing through roof. It's always been growing at a huge amount. I mean, right now, it's growing between 30% to 60%. And this is happening through all of the audio and the video feeds that are now permeating everything that we're doing from business. And so if you look at this, 80% of all of the content in the world is unstructured. And I don't mean just document files, like we used to talk about unstructured, but I really mean truly unstructured audio and video files as well as sensor data and others. And this content, you cannot make sense of it without AI. And this is really where AI is coming to the forefront of actually being useful to few companies today to actually make things -- to help them automate things, they couldn't automate before through all the interactions with cameras and microphones and other streams. And so this is really where Veritone is sitting, is taking all of this unstructured content, getting out of these amazing insights from this unstructured content and then running that through business processes for our customers and really changing the way they do business or interact with our customers. If you look at Veritone, we really have 3 main areas that we're focusing on today that will actually change how we do business and how the world does business. The first is intelligent process automation. That's currently growing at around 12% to 15% CAGR or growth rate. The next is pretty much where we started with AI as machine learning operations. So how do you actually take AI models and then scale them. And this is proof of what we've been doing with media, entertainment and other verticals of running our AI engines at scale, and we process massive amounts of content through our AI engines. And this is bringing this to the forefront and sharing it with our customers. The last one is really AI platform and developer tools. So again, how do you make it easy to collect all of the data that we actually have, drop it down to a data scientist machine, allow them to actually create different hypotheses and different models around it. And then actually build an AI model that actually solves their problem that they can prove and test it, and then deploy that back into our platform and then run it at scale. So these are the 3 top different market opportunities that we're actually targeting. And you can see the growth rates go from 12% to 15% all the way up to 80%. One of the things that I really love about aiWARE is it's really this notion of an operating system. And if you look at a normal operating system, a computational operating system, it really made it super easy to build applications that take advantage of hardware devices, such as hard drives and RAM and the floating chips that we used to have in it. We do the same thing, but we do it for cognition. And so what we've done is we've actually taken all these different cognitive categories. And we've actually standardized them. So that transcription is transcription, doesn't matter whether it's Amazon, Microsoft or Google, when you build your application on top of aiWARE transcription behaves the same. And you can actually swap out the models where the AI provider, as you like, without changing anything at all above the API tier. There's nothing that you have to change in the application. And this really makes it an awesome platform to actually build applications that leverage AI. And this is why it's so easy for us to actually build applications and turn all these additional business processes into solutions that involve AI. One of the things I want to share is really all the different solutions by industry. And so when you start thinking about aiWARE and all the different cognitive categories that are supported by aiWARE and you can see these as kind of rose, and I won't read them for you, but we support all of these and more. Look at all the different columns, which are the different business processes in these different verticals that we can actually target with AI and aiWARE. And so as you can see, there's a lot of mixing and matching of these services across. So that's not just, for example, if you look at law enforcement, I don't just need transcription. I need transcription, face, object and license plate, and then just kind of go through. So as you're building out these solutions, it's really the multiple cognitive categories as well that really drives it and provides us the power to help solve these problems in these different segments. One of the big things about an operating system is it has to run in a distributed manner where you would actually like to have the content. And so with aiWARE, you can actually deploy in the cloud today, where we run in all the major clouds, AWS, Azure, we also run in FedRAMP and GovCloud. You can actually run it in a kind of a mix and match mode where part of it's actually run in a cloud and then the actual processing of that content is actually running more on-prem, inside your environment or your data center. And then you can actually run on the entire stack, and this is coming soon, run the entire stack on your data center, where actually, it can connect to all the private data sources that you have, so that you can do better integrations with all these different business processes that you have. Thank you for taking the time for letting me share aiWARE with you. I'm very passionate about what we're doing. I'd like to now jump into a couple of testimonials about how our customers and partners are leveraging AiWARE and it's changing their life.

We initially approached Veritone with just going after transcription of all of our assets as well as facial recognition. But I think having the flexibility to layer on additional cognitive services in the future just gives us a lot more chance to grow our assets that are already in the system. So if our partnership team approaches us and says, "Hey, we want to find every Coca-Cola logo that was used in the 1980s with a couple of clicks?" We can seemingly pull up all that information and run logo recognition or graphic recognition on our content. So I think having the flexibility to layer on additional metadata in the future is huge for what we want to do. Additionally, Veritone brings us new opportunities in what they're developing and lets us add that additional flexibility to our workflows. And so one thing that stands out there is with automate studio-based workflows. They were already able to take the transcriptions that we had generated off our content and be able to help us generate SRT files and caption files for our videos. And captions are of growing importance to our need as we put out videos on social media, where fans might not be engaging with audio, but they want to watch captions. Additionally, when we're translating between multiple languages, captions play an integral role in our content. And we're just excited with where things are going to grow in the future with the platform.

As an analytic consultant, I'm frequently asked about machine learning and AI. And with Veritone, I never thought that I could do AI seriously without coding. I now have all the answers in one place for any structure on a structured beta sits, and it puts AI and ML at your fingertips. There is no coding required, but you can, if you wish, and it also give access to many AI engines. So you can get insights from video, audio, images, text and data, all running within a simple workflow. Basically allowing us to get unstructured data like video and audio, which for some of our clients represent the 80% of their data assets and convert them to actionable insights. And we were unable to do that prior to discovering this tools, absolutely. With the ability to access a lot of engines, but I don't need to have an individual contract with any of those, you don't need to understand and learn how each one works because using aiWARE, no matter which one I need to use or I want to use, for me, it's the same data preparation and format. I can offer all of the available AI engines in the market under one shop.

Hello there. My name is Corey Hill. I am a Senior Solutions Engineer at Veritone, and I'll be talking to you today about Veritone Interaction Analytics for conversational intelligence. And when we talk about Interaction Analytics what we're really talking about is enabling the enterprise at scale to better understand all of the touch points in their customer journey. Analyzing the sheer amount of interaction data that doesn't fit into a database is not cost or really time effective. So as a call center manager, how do I actually review the hundreds of hours of conversations I've recorded today for quality and assurance training purposes? And on same sense, as a brand manager and the marketing department, how do I review the hundreds of hours worth of video content, memes and images that get posted to social media? With the proliferation of Web 2.0, it's getting next to impossible for any brand manager to manage that sheer volume of data and derive good insights and analysis from that. And so with Veritone what we're actually doing is enabling you to extract, learn and visualize information based on these cognitive actions that we're referring to as interaction analytics. So that consists of things like transcription, so taking speech to text, recording sentiment analysis as well as using computer vision to detect logos, product images and customers through social media posts as well. So what does Veritone Interaction Analytics actually look like? As Interaction Analytics covers a broad spectrum of touch points across your customers' journey. We've actually decided to focus on 3 specific use cases that are really going to help drive change within organizations. So for example, from a contact center perspective, knowing who is speaking, when and how they're feeling is incredibly powerful tool to help drive the conversation in the right direction, reduce time to close as well as gauge and kind of really understand how your customer is feeling during that conversation. From a contact center compliance perspective, what we're trying to do is enable organizations to scale their operations without sacrificing customer experience or their regulatory standing for all interactions across the organization. Last but not least, social media, with the proliferation of Web 2.0, it's next to impossible to understand and analyze all of the data that's going across social media. And so what we're enabling you to do is take a really good understanding and narrow down what matters and provide actionable insights on that. So maybe you want to detect which products are showing up in images, getting the sentiment of the product reviews and getting a transcript of that on YouTube, for example, that's how we're enabling the marketing department to actually find insights, leveraging our social media preconfigured solution. As a core component of our democratization of the AI revolution, we've partnered with Alteryx, which is one of many tools that enables customers to blend and visualize data across data silos within their organization. And essentially, what we're providing is access to the 80% of unstructured data that exists in any customers given data universe today. And essentially, we're adding a jet pack of AI to their Alteryx workflows by enabling them to send that unstructured data, such as media, so files of video, audio recordings, images, IoT sensor data, et cetera, to actually start to ingest that information and get analytics on it, but then most importantly, correlate those outcomes and results to the business data that they have in their current structured database. So it's really about kind of blending the results from aiWARE with the information that is already in their customer databases to drive visualizations and ultimately decisions. Now let's hear what our partners have to say.

Customers turn to Alteryx to solve business problems with analytics and many organizations want to use data from unstructured sources, like audio, video and images, to drive breakthrough insights and automate processes with analytics. The Alteryx-Veritone technology partnership helps our mutual customers use both structured and unstructured data sources, in the Alteryx Analytic Automation platform, to develop new spatial, diagnostic and predictive insights and actionable intelligence that drive breakthrough outcomes in the business. Our shared customers are excited to use the combined offering of Veritone's aiWARE building blocks for ingesting unstructured data, within Alteryx designers, drag and drop analytics interface to help them automate every step of analytics, including data prep, blending, reporting, predictive analytics and data science. As the Alteryx-Veritone shared solution continues to mature, I'm excited about quick start offerings that help mutual customers immediately realize the value of this technology partnership, including starter kits that address popular use cases, like analyzing contact center recordings to ensure compliance or analyzing social media to develop insights on customer sentiment.

Hello, everyone. My name is Lance Gorji. I'm the Senior Director of Product Management here at Veritone for our aiWARE tooling group. I'd like to tell you a little bit about our newest labs project, Veritone Clarity. There are thousands of ML models in the marketplace today, with many claiming to outperform the competition. However, many of those models fall short of the expectation due to poor model performance or unintended consequences and inaccurate AI decisions. This can lead to problems such as text extraction models producing inaccurate output that requires correction. Facial recognition models, failing to correct and identify a specific race or even job seeker evaluation models that perform specifically for male candidates as opposed to female candidates. We'd like to answer these questions and more with our Veritone Clarity product. The Veritone Clarity product helps build visibility into AI model behavior and performance and helps you select the best AI model for the job, and it helps you detect drift and it helps you correct that drift. It helps you build that trust and explainability within each individual machine learning model, and it does so by providing a rich feature set of analytics that you can share with your customers, which will help reduce risk across your entire organization. You can easily evaluate, compare and monitor performance, cost, speed and behavior across all of your machine learning models and your competition, whether they're homegrown, third-party or aiWARE driven models. And next, I'd like to hand it off to Cory Hardman, who will tell you a little bit about our Verify project, which is another Lab script.

Thanks, Lance. Hi, everyone. I'm Cory Hardman. I'm a principal engineer at Veritone, and I've been leading the Verify product. We at Veritone looked across the industry and have come to conclusion that modern authentication is failing. It is failing users with the ever-increasing password rules and users having a hard time with remembering all these passwords. And on top of that, the passwords are still being compromised, and it's failing companies, with all the account takeovers and the worst customer experience. It's not to say that the industry isn't trying to fix this with things like 2-factor authentication. But when you look at 2-factor authentication, it is no more secure against phishing attacks than where we were with passwords. There are solutions like FIDO, but FIDO requires hundreds of dollars worth of special hardware tokens and a very complex user journey. The world needs a realistic alternative, one that is passwordless and authenticates who you are, not what you know or what you have. And ultimately, we should be providing real proof that this is the expected user that is using the device, not verifying the device itself. We at Veritone Labs have created a smart biometric single sign-on that looks at multiple cognitive off methods simultaneously. Users love this because it's fast, easy to use and their identity is finally in their control. App developers want to integrate it because it's secure, easy sign-ups for the user, and it reduces churn of users for simple things like forgotten passwords. The Verify team has been hard at work, and we're excited to show you how it works today. A user when presented will select log-in with Verify. The next question, which is, what you typically would see. They select their e-mail address that they're trying to log in as. What you won't see though is as they select it, we will not be prompting them for typing in a onetime code or their password even. Instead, we're going to show the user that we know who they are by verifying them. So let's go through it. Verify even shout map them. As you can see, we successfully logged in. During that log in experience, server side, we were considering numerous factors, such as the face print, the voice print and some lightness challenges like were my eyes naturally moving? Were my lips moving alongside the words that I was saying and ensuring that the user actually spoke the challenge phrase. All these things together, plus many behavioral analysis and strong device binding factors are all put together in our cognitive engines to decide a confidence score of how likely it is to be me that is logging in. The key thing is that those factors will keep improving over time, but the user experience will remain the same.

Hi, everyone. My name is Sean McEvoy, Senior Vice President at Veritone, and I'm here to talk about Energy. So our mission at Veritone is to accelerate the world's transition to a more resilient, reliable and sustainable grid. Whereas, most of you may know, this hasn't been an easy road for utilities, it's a fact it's been very challenging. And some of those challenges, as you can see on the screen here today include, for example, weather impact. And we've all seen in the last couple of years, the impact that weather has had for example, in California, through extreme heat, which has caused many disruptions and outages; also in Texas through the extreme cold, where it's also brought down the grid in Texas. And even like on something as simple as cloud cover can also impact renewable energy such as solar. And this is causing the utilities to incur and have in place large amounts of spinning reserves to fill those gaps caused by impacts from weather, for example. And these spinning reserves are typically diesel, our gas turbines, which are not very clean energy friendly. In addition to that, they're very expensive. And with like the Veritone Software Solutions, we are able to not only predict when extreme events are going to occur, but through our predictive controllers and automated decisioning, we can plan ahead in advance to keep the grids like more self-sufficient and minimize disruptions, minimize costs while at the same time, reducing carbon footprint. So let's take a look at some of Veritone's energy management solutions. In our solution set, we have 5 core modules; the forecaster, controller, optimizer, arbitrage and simulator. We have 28 patents issued against this software, and all ones in real-time powered by the Veritone aiWARE platform. The forecaster is able to predict supply/demand and price. Our controllers, which are predictive controllers, can control all of the assets on the grid, synchronizing them and providing optimization such as flow control and also optimization, such as optimal supply mix. We also have an arbitrage solution that allows independent power providers to be able to trade energy on the various wholesale markets throughout the U.S. And we also developed solutions in a rapid fashion using our simulator, which is built on top of our aiWARE platform. Our strategic energy plan includes 3 phases. In Phase 1, which has already started, we aim to become the platform standard for the entire energy industry, providing optimization and predictive control solutions to the utilities and microgrid providers. In Phase 2, we will enable the autonomous energy grid by integrating our predictive edge controllers with grids devices, such as inverters, transformers and battery storage for optimal performance. In Phase 3, we will manage and supply Veritone Energy to the wholesale markets as well as the interconnected grid using our predictive controllers and arbitrage solutions. This 3-phase approach will provide Veritone with exponential growth moving forward. One of the key topics of today is the autonomous energy grid. And when we think of an autonomous grid, we think of interconnected microgrids, all supplying energy interconnected between each other, able to both sustain themselves in terms of an extreme event, but also to help destress the entire grid in case of an extreme event, where we can plan and work with these microgrids to be able to automatically island themselves in case of an extreme event, and then we can also synchronize these microgrids back to the main grid when everything normalizes. So thank you, everyone. Now I'd like to share some testimonials from our partners.

Veritone, I think, is kind of at the leading edge in looking some of the things for AI and ML around energy. And they just came out with the grid, the grid reliability device, where they can take AI technology embedded on a lot of our green technology capabilities, such as solar, wind, battery power, EV. I mean, we're moving to electric vehicles. But the ability to now have a software capability embedded in all of the so that you can level the playing field and looking at energy specifically, level the playing field using -- taking all of these disparate sources, being able to manage them and if you have a severe weather impact or you have a disaster, another manmade or otherwise disaster impact, where you can then micromanage and say, "Solar is down, we need to move to wind or we need to move to battery, but we need to continue the provisions of energy," which is kind of a core basic need for populations, you can avoid catastrophes like Katrina or Maria or Sandy in communities, and you can have a continuity of effort. And that is going to be a game-changer. And not only game-changer here in the United States, but globally.

What we're doing right now is we're working very closely with Veritone, proposals to the federal government, mainly the Department of Defense side, dealing with smart buildings, microgrids, those are the technologies that are being widely demonstrated and deployed. We're using Veritone's Energy solution to build on advancements that we've been making to approve optimizing, synchronizing and automating energy feeds and sensors. And we're finding Veritone technology as a key enabler for these proposals as we're going forward with these opportunities. The system learns over time about the sources of energy production that feeds into the system. So over time, it grows more and more efficient without the dependencies of a human in the loop. Because the human is not in the loop, you can really get to a higher level of optimization. It's really quite brilliant when you get down to it. It's a phenomenal capability. And it offers that clear technical differentiation for large and small scale applications that allow us to: one, improve operability; two, increase efficiency; three, lower costs; and four, it's done with less dependency on human controls. So I think you'll see more demand coming out of the Department of Defense and the federal agencies for energy, security solutions that deliver energy at lower costs. It's not unreasonable or unrealistic to think that DoD installations of the future will have advanced technology microgrids. In fact, I believe that's where they're going to have to go.

Since inception, has successfully developed 2.7 gigawatts of wind and solar throughout the U.S. and Mexico valued at more than $4 billion. In addition, I've personally been involved with the development and construction of another 2 gigawatts of wind, 700 megawatts of PV solar and 1 of the largest energy storage projects in Florida. We believe Veritone Energy and its solutions will help bring significant value to the renewable energy space because of their AI technology and solutions. We at Amshore, as a developer, see Veritone and its value as a discriminator in a very competitive market. Renewable energy technologies like wind and solar are intermittent. The variability of these technologies creates its own problems, power fluctuations from cloud cover or constantly changing wins, and requires utilities to carry larger quantities of spending reserves. this can result in significant cost to the utilities. The Veritone AI technology helps predict what the renewable asset or a microgrid will experience during the next window of time. In knowing the future, the Veritone Optimizer then provides direction to the asset or system. The Veritone technology will enable the customer to maximize its revenue, minimize operational costs while prolonging the health of the assets. For example, a customer is implementing the Veritone system with a project that includes solar and storage. The AI platform allows the customer to mitigate the variability of the cloud cover and its impacts to the project solar output and operate the battery storage to reduce the cost of spending reserves, while prolonging the life of the battery. The Veritone AI platform is a key to expanding the implementation of renewable energy as a primary source of power. The first movers of the Veritone platform will have a distinct advantage over its competitors by optimizing their operations to, again, maximize revenue and reduce costs. Without an AI predictive control system, renewable energy will struggle to be reliable and economic option for future power systems.

All right. Quite a lot in there. We're having another Q&A session here. As the questions are coming in, I'm going to kind of throw a general question to Al here. You talked about the deployment flexibility and how that's been a big push for us, and Chad had mentioned it in the prior one. But maybe you can speak to the progress we've made in terms of the reliability and in terms of the performance, and what that means in each of these use cases to have that combination of reliability plus deployment flexibility.

Absolutely. So we talk a lot about deployments and how we actually deploy our software, and we can run it anywhere that you actually wanted to run, including on the devices that Chad and Jon had talked about. Part of what comes with that is really the ability to make sure, even if you can run it there, it doesn't matter if -- you can't run there if it doesn't work. And so we've been focusing a lot of the efforts over the last 4.5 years that I've been here on making sure that our entire technology stack just works. And right now, we are typically running for our SaaS environments, around 100% SLA for all of our API calls that are hitting our stack. And so that focus on reliability has made it super easy to actually deploy our technologies and leverage by our customers and partners.

That's great. First question coming in on the energy front from Nick up at Craig-Hallum. Sean, can you maybe go through the engagement level that you have currently on the energy front in terms of the customers, maybe put it in terms of scope of types of customers as well as just the number of people that you're interacting with?

Sure. Well, we're under NDA with every one of them, I can't kind of divulge names. I would say that major implementation currently wrapping up with a major utility in the southeast, that's around solar predictability, solar inverter control, frequency control, voltage control, which is all about improving the quality of the energy going into the grid as well as battery storage management, and these are like batteries that are like $20 million each. So this is a very sizable project, high implications. But what it allows to do is allows us to this utility to reduce a lot of their spinning reserve costs and also improve the quality and reliability into the grid. The pipeline that we have today spans across like multiple types on the, what we call, front-of-meter meter side, and you have various types of utilities, such as the investor owned utility. These are the big regulated utilities. We also work with -- in our pipeline, the municipals and also the co-ops. And then what we're seeing like a rapid growth is in behind the meter, where we're working with developers, independent power producers as they develop microgrids and virtual power plants. And in addition to that, we also have the manufacturers of the hardware that sits on the grid, such as transformers and inverters and battery storage that they need intelligent control, they need forecasting, they need predictability. And all of these solutions sit on aiWARE they can leverage all of these and also imaging capabilities as well as we predict out kind of weather and we're using thermal imaging tech cameras and we're ingesting all that content into aiWARE to get better predictions around weather, which impacts renewables. So a broad range of customers both that are on the kind of consumer side of the house and also on the production side.

Great. It's really helpful, Sean. Michael Latimore is asking kind of a basic question from -- not basic, it's actually fairly complicated. Al, maybe you want to take a stab at this. But he's asking, what is the unstructured data that we're getting as it pertains to energy? And how is that being processed by aiWARE?

Yes. So we're getting a lot of unstructured data across our stack, and not just in energy, but it's really the camera feeds for energy, it's also the infrared cameras that are pointing at the sky to give a better intelligence of what's actually happening from a radiation perspective. It's the audio feeds that we're getting from different systems as well as unstructured documents. And so we take all of that and run it through various different engines. So we have, for example, NLP, text analysis or sentiment analysis on that one. We have, as you saw, on the GLC side, we have face recognition, object detection, head detection, logo detection. We have transcription, translation. So we have a lot of different AI services and cognitive categories that really supports taking all of this unstructured data and turning it back into something that our customers can use and tools and systems such as our energy system can also use.

Let me jump on that one really quickly. So with regards to energy specifically, I mean what we're interfacing with is basically an on-prem deploy of aiWARE where the system ingests data from the PBE arrays themselves, which includes the power output, voltage and other types of DC data coming from the arrays as well as the inverters, transformers as well as the battery, which includes things like voltage, storage, temperature, et cetera. All of these factors not only point to the health of the equipment itself, but also how they interface in terms of the input and output of energy between those disparate devices on the network. And so this is an extremely complex distributed AI problem. Equivalent to almost the traveling salesperson problem back in college. And by deploying our proprietary CDI technology to the edge, it's literally learning how these systems work together as well as individually and creating optimal solutions on a sub-second basis and reprogramming the grid. What I found really interesting when we kind of entered this market is we had all heard for a long time growing up that we had -- we were in the process of deploying an "intelligent grid". Sean, can you comment on -- comment for us just how intelligent the grid is in its current state and what it is that we're bringing to the table?

Well, a lot of the grid today is manually operated. So you have like large groups of staff like pressing buttons. When things get too hot like a piece of equipment, let's switch that over to another one. There's no predictability, in most cases, around solar output, such that these utilities have to have large amounts of like spinning reserves, which are diesel or gas spinning reserves, which also cost a lot of money, so they're not seen like the benefits of -- not seeing the benefits of the renewables we're putting in. Not only that, the utilities are built around its legacy transmission infrastructure. And now you have distributed energy resources like wind and solar and hydro that are in different areas, all are trying to move energy in different directions. And that leads to congestion in the current grid. It also leads to curtailment of energy and also energy dumping. Now when you put like AI in this, AI is figuring out, for example, flow control, what is the best optimal path to route that energy to -- in the fastest way possible to the destination without avoiding like all types of congestion, so you're not losing any energy. Also, at the same time, because we're looking now across multiple types of energy producers, such as your solar, it can be battery storage, it can be hydro, when you have demand or a spike in demand, you want to know what is the best supply mix that you want to give to the grid at the best optimal cost so you can pass those cost savings on to your rate payers. And our models, AI models, are updating in real time such that we know at any -- like in a millisecond what is the best optimal supply mix. And what I mean by that is like on a per second basis, we are determining, to meet that demand on the grid, this second, you should take, for example, 20% from solar, 50% from your battery storage, 20% from your hydro, taking into account rules such as NERC and FERC rules for quality but also cost as well. So you're doing it on a cost priority basis.

Chad, I want -- a follow-up question here from Chad Peterson. As it pertains to CDI, you mentioned the distributed nature of AI. He's asking, can you describe CDI in layman's terms, first of all? And then what is the extensibility of it? And can it move into other markets such as health care?

Yes, absolutely. So in layman's terms, CDI is a distributed intelligent infrastructure that runs principally on the edge of a network, but doesn't necessarily have to. With very complex systems that are changing in real time, you historically don't have the ability to gather massive amounts of data and train a classic AI model like a deep learning neural network on that system because of the variability of the environment. Deep learning neural networks and traditional networks need a very stable data set. For example, if I wanted to teach a system how to transcribe the language or recognize faces, I would gather a set of information about faces or language and train a model for a period of time, and that model will then be deployed into production. When you think about energy grids or distributed systems that have high variability, what you're seeing in there is the network itself is changing continuously. And you're constantly adding new systems to that network, including new customers and new suppliers in the IPP side. And so you need an infrastructure and a technology that can actually build models on the edge without the necessity for a data scientist to be plugged into that system. So this is what CDI does at its core. To break it down a little bit further from a mathematical standpoint for those of you, CDI actually looks at an extremely complex system and figures out mathematically the rules and forces that govern that system and are then able to apply human-based rules, both soft and hard rules against that model in conjunction with what we're seeing. And the beautiful part about this is it's ideally designed for large distributed systems at the edge. So this would include things like 5G network routing. It would include understanding large medical imaging systems. It obviously applies to energy routing and transportation. So we're just scratching the surface with regards to what we're doing with CDI technology and energy, but we're staying very laser-focused at this point in time and only deploying the technology there to ensure success with that, obviously, massive market.

It's great. We've got a ton to go through today, but I want to get one more question in this segment. This is a bit of a softball for you, Sean. A question coming from Mike McCalla. He's asking, can you provide some color on the size of the subscription cost for our energy solutions as it pertains to the utilities? And will that cost be based upon the size of the utility? Or is it based on something else? And given what we know from our pilot project, can you give some detail or some scope in terms of the size of the opportunity as you see it today?

Sure. So typically, when you're dealing with a utility, you're dealing with a mass deployment. We have seat licenses for all of our software, for our controllers, for our forecasters, optimizers. They are typically controlling individual assets on the grid. So you have a seat license typically per asset. And then like they all have like individual pricing. And it's typically utilities are like enterprise, high pricing. They haven't moved more to like the typical models. And so we're charging either on like it's an annual kind of subscription license. It can be also a term license, and someone may even want a perpetual license. But -- and they all come down to the maintenance and support over multiple years, 24/7 because they are mission-critical systems. And then on top of that, the size of the projects are all 7-figure dollar projects because of typically what you're dealing with, the megawatts of energy that is flowing through the systems. The solar arrays, for example, you've got our solar firm with like 120, 140 inverters, multiple batteries, multiple battery inverters, multiple transformers, multiple solar inverters as well. So a lot of seat licenses for a deployment even just across solar farm -- not taken -- you can then imagine like what it looks like on wind turbines, where you have hundreds of wind turbines with similar-type pricing and licenses.

That's great. Absolutely awesome. We got a ton coming on throughout the rest of this morning. We're going to take a quick intermission here. It's 5 minutes, you'll see a countdown on the screen. There's also going to be a couple of sizzle reels for those of you that are just going to stay put. But we'll be back in 5 minutes from now. See you soon. [Break]

Okay. Welcome back. Thank you, everyone. I want to give a shout out to all the presenters and all the testimonials and all the customers that have engaged in this event thus far. But the second half, frankly, I think, is going to be as exciting, if not more exciting. With that, I'm really excited to introduce our media and entertainment business with Sean King, who's going to kick things off for the second half.

Depending on where you are today, good morning, good afternoon and good evening, everyone. I'm Sean King, Executive Vice President of Veritone One, and I would like to thank you for joining and staying on this virtual tour of our business. Today, I'll be sharing an update on Veritone Media. In this section, we'll share the market landscape, our target markets and our AI-powered portfolio of products and intelligence services. Throughout the presentation, I'll be joined by a few colleagues. Jeanette Kadow, Regional Sales Director; David Tetley, Customer Success Manager; and Lauren Cowles, Director of Program Management. And lastly, I'm proud to share the screen with our President and Co-Founder, Ryan Steelberg, who will be sharing an exciting announcement. Last year, the media and entertainment and ad industry absorbed a historic shock to the system. It disrupted long-standing business models, accelerated existing trends and created new opportunities. Most importantly, COVID-19 drastically changed consumer behaviors and their insatiable demand for content at home and online. This topic, we'll touch upon later. The decline in 2020 has expected a big bounce back in 2021 and beyond for both the media and entertainment and media ad spend landscapes. The media and entertainment industry is expected to grow at a 2.8% CAGR, roughly similar to the long-term trend. Global ad spend is forecasted to grow at 15% to $747 billion this year with continued steady growth onward. As we dig into the global media ad spend projections, we're noticing an interesting trend as it relates to our own media and entertainment business. According to our recent advertiser perception studies, marketers are putting 20% of their digital ad spends into branded content and influencer marketing. For the first time ever, branded content and influencer marketing will account for a larger portion of digital marketing budgets than any other category. This includes paid search, display, paid social and digital video. Further, more than half of the survey respondents that use branded content and influencers said doing so is more critical than it was a year earlier. And about 1/3 of advertisers said they boosted spend on branded content. So although there are still challenges for media and entertainment companies as we move beyond the pandemic, the digital migration that is pull forward will generate significant opportunities for all of our segments. Veritone is uniquely positioned to help our partners and customers capitalize on this opportunity through our core technology, aiWARE, and the applications and intelligent solutions which it powers. Speaking of segments, these are primary targets for our media business. But more importantly, we want to share our expansive roster of top-tier clients across both our applications and services that we offer in these segments. It includes the largest audio company in the U.S., the largest sports network, the major broadcasting networks, the #1 ride-hailing app, the #1 B2B social network and much more. How do we support thousands of customers across these segments? First, it's important to note that Veritone's suite of products and solutions are all built on aiWARE, the world's first operating system for artificial intelligence. These products help our partners and customers assess in near real time, everything they're working towards from a data, content or advertising perspective. In addition, our Intelligent Services divisions from our advertising agency, Veritone One, to our content licensing business and revenue solution, VeriAds, are all built and powered by aiWARE. Let's take a moment and dive into each. As referenced earlier in these presentations, 80% of the world's data today is unstructured, trapped within images, audio and videos, and it's growing at 30% to 60% per year. Applying AI to these media forms to structure this data enables anyone to determine what happened and when within this content. Discovery gives radio, television, podcasters and other content creators tools to discover large amounts of visual and audio data sets for robust delivery reporting in near real time. Powered by aiWARE, Discovery identifies everything that set or seen in any content, including words, faces, logos and even the sentiment of the content, bringing structure to this data in a captured media that was previously unstructured. This transformation of audio and video content enables broadcasters to quickly find and view time correlated playback of media and share this information with advertisers, sponsors or partners. It enables intelligent ad and content discovery and has widely been adopted by top media groups in the U.S., currently serving over 1,500 customers in 4 countries and processes 1 million hours of media per month. Up next, we'll show a customer testimonial from one of our value discovery customers, Westwood One.

Veritone has really been fantastic for Westwood One because the reason -- the main reason that we use the Veritone platform is to find 10-second custom features that we're sending the clients. So when we need to weed through an NFL game or a collegiate basketball game or a golf match or anything that is sports-related, live and game content. We're able to search by keywords. We're able to hone in on the targeted message, cut it and send it off. And a new area that we've been working with Veritone has also been the podcast platform. So when you have a podcast, that's an hour or 2 hours or even 45 minutes, when we're looking for that 60-second bit of content to send to the client, Veritone really helps Westwood One solve for a targeted approach to make sure that we get the client what they need. So Veritone's really helped Westwood One, as I've mentioned, cut down on the time that it takes to find airchecks for sponsors across sports and podcast. And what that really allows us to do is optimize our time, whether it is customer support or spending more time on actual day-to-day sales. The Veritone platform specifically, but the chance to download an aircheck from the actual platform, it really just makes it so much easier for our marketing team because then they're able to implement the airchecks that we're specifically sending them into presentations to share with clients. So yes, before, airchecks would come across from our traffic team, they still do for a number of syndicated shows. It's not to say that Veritone has completely wiped out our traffic department by any means. But it's really helped the team optimize where we could spend our time. And if someone's focused on Veritone looking for sports sponsorship specifically, the way that we're able to trim it, download it and share it with our marketing team, just makes it for a smoother, faster process, so we could focus on more of our day-to-day business or get clients what they need even quicker than we've ever been able to before. So I really think that Veritone, Westwood One specifically, again, I can't hone on this enough or emphasize this enough, I should say. But Veritone has been so instrumental in making sure that we could target clips, whether it's 10 seconds, 20 seconds, 30 seconds of audio, from a much larger audio file upwards of hours or days when you're looking at a golf tournament, I really think Veritone from our use specifically is only going to continue to expand. And as I mentioned, we've already started to use Veritone in the podcast space. And a lot of the audio industry as a whole is looking towards podcasting. Just from a content perspective, is a new way that people are consuming media. I think that Veritone is going to continue to play an instrumental role in being able to help companies like Westwood One, really hone in and focus on messaging, whether we use Veritone to get clients, airchecks to solidify that an ad ran to make sure that it ran correctly, to give them time stamps when it ran. I think that Veritone is going to continue to be a really instrumental tool for any audio company, but Westwood One specifically in delivering new assets as audio continues to grow. So it isn't just play-by-play sports anymore. Podcasting, as I mentioned, which has been a huge area of growth for us on every single side that you can imagine. So they have Veritone sort of already set in stone in a well-oiled machine, just makes the whole process so much easier on our end as we look and focus on newer ends of the medium. So I think that Veritone will continue to play a huge role from a live sports perspective, from podcasts specifically. And then as we tap into syndicated shows and working with personalities, the tools that Veritone provides Westwood One really make it a lot easier to optimize and make sure our clients have what they need.

Thank you, Blair. Now let's turn our attention to Veritone Attribute. Disruptions due to the pandemic caused a lot of ad dollars to divert towards performance-driven media, specifically digital. As advertisers come back to broadcast media, they expect the same metrics of success to prove ad advocacy and ROI. As such, there's a need to capture and visualize this data. Veritone Attribute helps broadcasters and content creators, measure and correlate on-air placements and in-content mentions on their online website, page and search lift. Using aiWARE, Attribute comes with a robust reporting for optimization strategies, campaign analysis and management and period-over-period analyses to maximize ROI. This is changing the game for these traditional channels who have historically struggled validating campaign performance with metrics of success for their advertisers. Currently, Attribute tracks 6,000 ad campaigns on approximately 1,200 stations globally. Up next, we'll hear customer testimonial from America's #1 audio company, iHeartMedia, followed by a short demo of how discovery and attribute are providing near real-time insights for broadcasters from Jeanette Kadow.

Well, first off, thanks for letting me join today. It's a real pleasure for us. And here at iHeart, we're focused on solving 2 challenges. The first being, we need a solution that allows us to provide attribution solutions for broadcast and audio, metrics that our clients are used to seeing for digital campaigns, and they expect from audio. And the second thing we're focused on is finding a way to demonstrate to advertisers, whether they're new to radio or skeptical, that radio works and drives immediate trackable results. iHeart's leading methodology was incorporated into the Veritone platform. And the output of the solution allows us to provide clients with actionable insights to optimize their broadcast campaigns in near real time. At iHeart, we feel strongly that attribution solutions are valuable at several levels. So the ability to isolate specific factors like a specific piece of creative day of the week or day part, these are all immensely beneficial. These checkpoints allow us to provide continuous improvement of the effectiveness of radio to our clients.

Hi. I'm Jeanette Kadow, regional sales director for Veritone. I'm excited to walk you through demos of our AI-enabled analytics and attribution products. [Presentation]

Thank you, Brent, for the kind words and Jeanette for the helpful demos. Now let's talk about Digital Media Hub, our aiWARE powered media management platform that is designed for content owners across TV, film, news, sports and other content producing entities. It simplifies the content creation with powerful search tools, which help anyone find in-store relevant media. With built-in metadata management and bulk editing capabilities, users can more easily manage the data of their media assets so they know exactly what they have in their archives. Digital Media Hub currently powers major content owners and content creators all over the world. With how fast content is being produced, content creators and rights holders now have a tool to easily access, search, manage, share and monetize their content internally with partners and with potential customers in this media. Now we'll share a few testimonials from our partners as well as a short demo from David Tetley.

The Veritone AI platform and the DMH platform that we use, it's helping us to understand our content and actually find the key moments within our content in a way that we can't do just with title level metadata. It is important for our business to be able to understand when we make a movie or a television show, what are those scenes where a particular actor appears, what are the scenes where there's a particular feeling or a motion, where do they go to New York City, when is it raining, when do the lovers kiss. These are the types of things that we need to find for requesters that are coming and looking for scenes from our movies because they want to put it into maybe a commercial, maybe into another television program, where the characters are watching TV, it needs to capture a certain feel or tone or mood. Or we're looking for content that we're trying to push out on to social channels to capture an audience and finding those clips that an audience is going to respond to and then hopefully drive them to purchase that title or rent that title down the road. Before we started working with time-based metadata and AI-generated metadata to index our content, our solutions to finding scenes was for a group of people that knew our library quite well to sit around in a room and try to exchange ideas and figure out if we could remember where a certain scene might exist within our catalog content. Now actually having the data that exists, a process that, A, was likely never going to find all the best moments; and B, was going to take a lot of time can now happen relatively quickly, where any user on our team or any of our partners can go on and look for something and find a moment or find many moments that are relative to the type of need that they have for that particular clip.

Our partnership with Veritone has grown over the last couple of years. What essentially started out as an opportunity to catalog and help us understand all of our assets that were in our Tape Legacy Project, over 17,000 assets sitting previously on the shelf, Veritone with their AI allowed us to understand so we could relive that footage and bring it back to our fans and make current again. Effective and efficient workflows for our producers and editors. But the other exciting part on the horizon to be able to play around with synthetic voice and MARVEL.ai for in the entertainment industry that is so rich in audio history, be it radio, broadcast or TV broadcast, even to Renault, our PA announcer. The opportunity to essentially play around and enhance the fan experience and bring in new voices that maybe our fans hadn't heard in a while or give our fans a different audible experience to something that enhances their entire experience of the game.

Hi, I'm David Tetley. I'm a senior customer success manager here at Veritone, and I'll be walking you through Digital Media Hub. DMH is an all purpose white label asset management platform. We integrate with a number of popular cloud storage providers to host all of your content from one place. Combined with AI or cognitive engines, DMH enables you to find, manage, share and deliver that content from a single easy-to-use tool. From the home page, you can use 1 of 2 built-in content highlight features to make discovery easy. Gallery review dynamically populates your content based on metadata tags, while highlight view allows tailored duration of constant categories. Alternatively, use the search bar to find a more specific content. Once you're on the search page, use the filters to the left to narrow down your results. You can also preview and download directly from here. But if you want more details, click on the asset to be brought to the asset details page. From here, you can view cognitive time line data powered by aiWARE, management data and asset information and download content directly. Once you found which you're looking for, curate that content collections to make public or share with your partners. Use the share features to send your content directly to other users or make the content public for anyone to view. From here, you can also make bulk metadata data updates to the assets within a collection or send the multiple assets anywhere from an ASP bucket to AI work processing. Once your users are ready to download, content delivery is a snap. Choose between 1 of 2 delivery models: packages, partners who can download freely; or our e-commerce feature, which allows your clients to purchase content directly from you via DMH. With Digital Media Hub, you can grant permission access to partners or clients, allowing them to view and download only what you want them to, control everything from content visibility to download access based on individual permissions to facilitate multiple types of access without having to leave the site. We're just scratching the surface of Digital Media Hubs' capabilities. It's a great tool that allows for secure but flexible access to your library and enhances your ability to not only find and review content but also share and deliver that content to your most important partners and clients. Thank you for your time. Back to you, Sean.

Thank you, Jason, for being a long-time customer of Veritone, and David for the informative demo. With intelligent services enhanced by aiWARE and its applications, Veritone One has become the world's first AI-powered performance audio and influencer agency that has harnessed the power of AI to measure and scale both traditional and emerging channels. Leveraging the power of aiWARE has enabled Veritone One to deliver unmatched insights that drive our media strategy, campaign messaging, reporting and analytics on behalf of our advertising partners, and more importantly, has allowed us to do this at scale. Because of the near real-time insights from our ad placements that aiWARE provides, Veritone One has scaled media that is traditionally very time consuming, such as podcast advertising, YouTube influencers and radio, but more specifically, native base placements across these channels. Let's hear from 2 of our advertising partners, DraftKings and Purple, who share some of their thoughts on Veritone One.

This is Josh Levin-Scherz, Senior Marketing Director at DraftKings, here today to talk about Veritone. Veritone has been an essential part of the DraftKings team for over 7 years. What stood out to me over that time is their emphasis and focus on precision, measurement and optimization. My team is in the dashboard almost every day looking at what our creative is doing and looking at what our competitive set is doing. Really excited to see what Veritone has in store for us today.

I'm James Brohamer, I'm Omnichannel Marketing Director at Purple. Purple is a sleep innovation company, most known for its mattresses, the Purple Mattress. When I first brought Veritone One, I was skeptical of podcast. I thought podcast was like influencers. I didn't know if they were working, and I didn't know who was working or why it was working. Since we brought Veritone One in, I'm able to go in and listen to every podcast. I'm able to vet every podcaster. Again, weekly reports on the podcast, both from a short-term and a long-term results perspective. Generally speaking, the podcast seemed to perform over a period of time. And depending on the podcaster, that time period differs and tone's AI technology platforms, they -- is able to present me with the data that I need to look at the podcast holistically and then layer that in with all the other audio buys that we're doing. I would highly recommend Veritone One as an agency, and I look forward to working with them in the future.

Thank you, Josh and James, for those positive words. Veritone licensing serves a critical role in the media and entertainment industry, with the ability to process media archives of past and current content while appending the necessary aiWARE-enabled metadata, which powers search and discovery. We help content owners to further monetize their media that they've created and ensure proper rights and clearances are in place to protect their content. With the vast archive of news, sports and historic footage, we give producers and creators the content they need to bring their projects to life using actual footage versus trying to recreate it. What better way is there to tell a story than using the real footage that can authentically transport the audience to that moment in time? But in order to find a single moment among millions of hours effectively and efficiently, you need a powerful AI operating system. Let's hear what our partners, CBS News and South China Morning Post have to say.

I'm Maggie Dakin, Director of Archive Sales for CBS News Archives. For over 90 years, CBS News has documented the most pivotal moments, both in our nation's history and across the globe. Over time, it became clear that these stories would hold immense value for generations to come. With that, the CBS News Archives is born. We have been partnering with Veritone licensing for the last decade to make our content available for story tellers, creating moving and informative films, documentaries, episodic series and podcasts. As we digitize our archives, Veritone's aiWARE provides a unique opportunity to solve one of the biggest hurdles in our business: search and discovery. As we look to our digitizing our entire archive, Veritone's technology will play a pivotal role in improving the licensing experience for creatives.

My name is Vincent Lee. I'm with the South China Morning Post here based in Hong Kong. We're really, really delighted with our partnership with Veritone, which has just established -- to showcase all of our great and compelling content that we have here coming from Hong Kong and China and from Asia to audiences and content -- to audiences around the world, specifically content creators in North America. To give you a little background, the South China Morning Post as one of the leading global news organizations here that reports on China and Asia for more than 117 years, and we've been a unique link between China and the rest of the world, telling what's going on this part -- the side of the globe. And we are very, very confident working with Veritone to bring this kind of compelling content through their amazing AI to the rest of the world.

Thank you, Maggie and Vincent, for your valuable partnerships. VeriAd is a full suite of revenue solutions that helps broadcasters, podcasters and content creators, monetize their content effectively and efficiently, generating incremental revenue. So they can focus on what they do best, creating great content. It also helps in the discovery process for this content, helping us identify the next big podcast, radio, social media personality or influencer and helps make them more viable for cash-based ad buys with aiWARE. By providing a revenue solution from handling clearance, ad verification and time-consuming trafficking of ad spots and revenue disbursements, VeriAds maximizes revenue opportunities and delivers a win-win for both creators and advertisers. This wraps up the Veritone Media section. We've spoken at length about how we're helping media companies and brands extract insights from their content and advertisements to help make better and more informed decisions. But another challenge they're facing is keeping up with the increasing content consumption habits of today's consumers. Now I'd like to pass it to Ryan Steelberg, Co-Founder and President of Veritone, to share some exciting news. Ryan?

Thank you, Sean, and congratulations to the media and entertainment group and all the BDUs for putting together such a great quarter, and we're really optimistic and excited about 2021 and beyond. So thank you, Sean. And now I'm so excited to introduce to you our latest innovation, MARVEL.ai. MARVEL.ai is a new solution that enables the AI synthetic content creation at scale, and what I mean by that is the ability to create hyper realistic and custom voices digitally generated through our AI models. Let's watch a short clip to help us visualize the power of MARVEL.ai and synthetic voice. [Presentation]

Before going on, let me explain what synthetic voice is. Synthetic voice or voice cloning is the creation and replication of someone's voice synthetically through AI and machine learning algorithms from audio samples. Very similar to how we use training data to train our other cognizant models such as natural language processing or object detection, we need lots of training data in someone's voice, whether that's a real individual or fictional character to produce enough quality high fidelity training data to create these hyperrealistic human-like synthetic voices. These new voice models will be supported and able to be driven by 2 modalities: text to speech and speech to speech. Text-to-speech functionality is where we use text input, it could be a written article, a document that we feed into the voice model and the output audio file is in that actors or character's exact voice. Speech to speech is actually using a voice actor, who then when they speak, is programmatically and automatically converted into the target voice model. Veritone is uniquely positioned to bring MARVEL.ai to the market and at scale. So why Veritone? It starts with our heritage and our subject matter expertise in not only digital media, but audio technologies. We actually mobilize and prosecute over 75,000 unique voice campaigns each and every month. Second, our technology. Built on aiWARE, MARVEL.ai provides a complete holistic end-to-end voice-as-a-service solution from data ingestion to training data preparation, to AI model creation and orchestration to ultimately producing the most highest quality, human-like hyperrealistic voices on the market. And then lastly, our ecosystem. Today, we work with over 2,000 media and entertainment entities all across the world, from major broadcasters, to major studios and networks. It's this ecosystem, it's this technology and it's this subject matter expertise that together create MARVEL.ai. Built on Veritone aiWARE, MARVEL.ai enables companies to leverage multiple best-of-breed cognitive engines and combining those with other cognitive capabilities such as translation, sentiment analysis, content classification and more. We help ensure they leverage the best solution possible for their specific needs and to create with the highest quality on the market. And now I'm going to turn it over to Lauren, who's going to give us a live demo of MARVEL.ai. Lauren?

I'm Lauren Cowles. Director of Program Management at Veritone, and I'm excited to showcase the power of MARVEL.ai, our exciting new voice-as-a-service solution. Let's log in to get started. MARVEL.ai is a complete end-to-end solution that allows you to create, manage, share and monetize professional quality synthetic voice. Let's start with voice creation. You can choose from over 200 existing voice models or you can create a custom authentic sounding voice using our managed services team. Here, we'll use text-to-speech using existing professional quality voice models. You can choose from a broad, diverse and growing marketplace of genders, languages and tones to create your voice content.

Hi, Ryan. I hope you're having a good day.

Creating a custom voice requires audio training data and Veritone's expertise at building AI models and voices that are suited to your specific use case. Here, we show 3 examples to showcase Ryan's custom voice. Let's click to hear Ryan's voice before any processing has been done.

Hi, this is Ryan. Your voice is the most important instrument of all.

Now let's hear his synthetic voice clip using text-to-speech processing, so you can get an idea of the before and after.

Hi, this is Ryan. Your voice is the most important instrument of all.

The last example is a female version of Ryan's voice based on his tone and style, using speech-to-speech processing.

Hi, this is Ryan. Your voice is the most important instrument of all.

These examples are just the beginning of all use cases, customers will be able to unlock with MARVEL.ai. From iconic voices from the past to those of today and everything in between, MARVEL.ai changes how you can create compelling audio content. As you start creating hundreds and thousands of voice clips, it can easily get unmanageable. The project tab helps you securely manage and organize your training data and voice clips into collections called projects. Each project allows you to preview clip recordings, download existing voice clips, or create new voice clips. As you are creating clips, efficient support of your downstream workflows is top of mind. Manual downloads and sharing are cumbersome and not as secure. MARVEL.ai enables custom workflows that allow you to programmatically send voice clips anywhere. With a native workflow designer, MARVEL.ai can create custom processes to retrieve and push voice clips and data from our to any destination. Plus, you'll have the ability to set permissions for sharing and protect your content with inaudible watermarks to ensure that only those who have granted access to can use your content. Furthermore, MARVEL.ai is built on aiWARE, Veritone's operating system for AI, and the user experience is consistent and connected to existing Veritone applications. This means you'll have access to evolving best-of-breed voice engines and be able to combine them with other cognitive capabilities like transcription, translation, sentiment analysis and more; two, deliver the highest quality voice content and make them easily searchable and discoverable. And now one of the most exciting things about MARVEL.ai, the voice marketplace. Here, voice talent from celebrities, on-air personalities, micro to macro influencers and more can add their voice models, exposing them to our vast fire network. Coupled with our extensive expertise in content licensing, this will open the door for talent to monetize their voice and get discovered for a variety of new opportunities. Thank you. And back to you, Ryan.

There are 2 important market dynamics that are helping drive demand and fuel the interest in MARVEL.ai. The first is the ever-increasing consumption of digital content, up over 200%. Additionally, the demand for branded content and influencer-based marketing for the first time surpassed search, display and social media advertising. MARVEL.ai is a perfect tool to take advantage of both those trends. MARVEL.ai will help democratize content production by making it cheaper, faster, easier with more high fidelity than anything we've seen on the market yet. There are so many industries that are going to be completely overhauled and revolutionized by programmatic content production. MARVEL.ai is bringing great value and efficiency opportunities for media companies. It will provide them with the ability to accelerate content production by converting text-based media into quality audio content. They'll enable them to programmatically transform and localize content across accents and more. It will also enable them to extend the reach and revenue opportunities of their voice talent and personalities. It will enable them to revitalize classic and iconic voices of the past from different talents and personalities. MARVEL.ai is particularly interesting for digital influencers and broadcasters. It provides them the ability to scale the capacity of their production to -- in creative and commercial projects. It allow them to create new and innovative revenue streams by the ability to produce more content that is hyper-personalized, localized for individuals. MARVEL.ai is bringing significant value and opportunity to brands and advertisers. It allow them to increase content output to keep up with advertisers' engagement and its audience. It'll prove advertisers response with more personalized content at scale. It will significantly reduce and shorten time lines to mass produce advertisements and marketing content. It will also ensure the consistency and continuity of the campaign's voice. MARVEL.ai is going to provide an incredible tool for government and public safety agencies. The ability to programmatically create and distribute content at scale to communicate with their -- our constituents or our populace. The ability to dynamically and programmatically change the language, accent or dialect, to be able to communicate to the entire populace at scale. And now I would like to introduce you to a voice you might recognize if you are a hockey fan or live in the Bay Area. For over 2 decades, Randy Hahn has been the voice of the San Jose Sharks, and can be heard on TV, radio, a Shark streaming service and video games.

I'm Randy Hahn, and like many other sports and TV personalities, I'm excited about MARVEL.ai. There's only so much time I can devote to endorsements in my role and my brand recognition is at its highest demand during a hockey season. And of course, that's when I have the least amount of time to support local businesses and charities due to my schedule. Veritone MARVEL.ai provides me with such a wide range of possibilities for my personal brand and endorsements because of its ease of use, minimal time commitment and control over the final voice file. I have spent countless hours in a cold dark studio doing live reads for video game voiceovers and marketing promotions. And with MARVEL.ai, this can now be done at scale and with my verified voice while I'm traveling with the team. It immediately changes the game for content producers, advertisers and influencer brands.

Thanks, Randy. As you heard, MARVEL.ai will not only enable Randy to scale his voice and brand, but it sounds like it may give him a slightly better quality of life by avoiding those dark, cold studios. MARVEL.ai is not just about generating incremental revenue or saving money. It is also about protecting the copyright and our trademark of someone's voice. The voice of a brand, the voice of an advertiser or the voice of an influencer is a very valuable asset and needs to be protected. Let me talk about our promise for good. Veritone upholds the highest standards of security and believe that transparency maintains that trust. We will always protect you, your voice identity and any IP you create with our technology. This includes the prohibiting the generation of and learning of unauthorized voice models, having controlled access to the voice models and also providing audible and inaudible fingerprint verification within the output of these models. Veritone will publish industry best practices and governance for synthetic content usage in public or commercial channels. Veritone is building a policy group, operators and buyers for self-regulation within commercial use across all industries. Veritone is an active member of the IAB and open voice network to develop global best practices for synthetic voice. MARVEL.ai leverages the best AI engines available today, both organic and inorganic to Veritone. Through our deep domain and industry expertise and with our focus on self-governance and protecting the voice IP, Veritone is uniquely positioned to bring synthetic content to commercial quality and scale. I would like to thank all of our partners and customers who supported us today. And with that, I want to turn it back over to Chad for a quick update.

Thank you, Ryan, and thank you all for being here with us today. I am Chad Steelberg's clone or Chad 2.0. I actually prefer the new and improved Chad. I have been listening and learning with you for the past several hours, and it's fascinating. I am literally a hint of what Veritone is about to release. But today, I'm simply something you can marvel at. But in 6 months, I will have learned enough to probably host this entire event. So I look forward to seeing and working with you all in the near future. Okay, legacy Chad, back to you.

Chad, you're muted.

I'm muted? I'm not muted.

You're back now.

Okay. Good. Sorry, guys. That was spectacular. I mean, what can I say, Ryan? Thanks for that. I was a bit surprised. But as you can plainly see, MARVEL.ai opens the door to truly all synthetic content, and we're just scratching the surface with our voice-as-a-service. And as I mentioned, I think in my top preamble remarks, human innovation has historically been limited by the number of people on the planet and the rate at which we can communicate. But it's only a matter of time before Veritone enables both humans and AI to seamlessly work together to solve some of the most challenging problems both today and tomorrow. So a great job on the launch today, you guys at MARVEL.ai team, and I look forward to seeing this grow in the future. Turn it back over to you, Brian, and media team for Q&A.

Awesome. I mean, gosh, ton of information, great testimonials, Randy Hahn, those of us in the Bay Area that's -- I don't know, milk and cookies. Great stuff. We -- questions are coming in here in the background. But maybe I'll kick it off with you, Ryan. You and Chad have been thinking about this opportunity to bring synthetic content creation and the personalization of content for a while now. And MARVEL is just the start. Can you maybe walk through how you see things playing out with the launch today? And ultimately, as we move forward and get to maybe a new and improved Chad 2.0?

I just -- God, I hope so. But this -- I think we -- this goes back a long time that it shows truly the power of aiWARE. And I keep going back to that because when we've been able to ingest, organize and index all of this audio and content, which we've been doing for years, right, and we've been able to generate revenue and provide great value in media and entertainment companies, it's also giving us sort of laying the groundwork for us to have great organized training data and orchestration of AI models to then be able to start producing both augmented content and what we're sort of presenting here today, fully synthetic programmatic content, okay? And we see that -- and there's a lot of -- and I'm going to do an analogy ironically to our energy group, okay? Historically, if you look at these legacy grids, you have energy input functions, right, or supply from a few sources. And then you have these rigid grids that then distribute, right, in kind of a circuit nonpacket-based environment to households and different distributions. That's how the media industry was forever, right? You had 4 or 5 networks that controlled all the content. It was distributed through very discrete lines of distribution, linear television and whatnot. Today, you have literally millions and millions of not just media groups that are producing content because now we, as individuals, are producing the content, right? So the amount of supply that's coming on to the ecosystem and the opportunity and the sheer volume of consumers is affording us to say, what is the best content to deliver to Chad or anybody at the right exact time? That content may have not been created yet, meaning we're going to create it dynamically on the fly to deliver the right experience, the right content to the right user. And that's just -- and MARVEL.ai, I think, is just the beginning part of that. We're focusing on voice first. But obviously, as you sell little Ts, we're going to be expanding the capabilities and functionality of MARVEL.ai to approach the video avatar as well.

That's awesome. And in your presentation, Ryan, you talked about the ecosystem that we have as a company. And I think that's a great segue for Sean to maybe talk about how aiWARE enables Veritone One to deliver unique and, frankly, superior performance for our brands, especially for the podcasters and the influencers. Can you maybe talk to the scale of our ecosystem, Sean? And why continually, you win?

Yes, I'd be happy to, Brian. What really makes Veritone One special and what allows us to be able to do what we do at the scale that we do. And Ryan touched on it earlier, we're on average, working on about 75,000 unique integrations a month. So I'll say that again, 75,000 a month. And this is a very time-consuming process because we're relying on people to lend their voices on behalf of a brand in their programming. And this is the whole broad spectrum of macro and micro podcasters, ones that have millions of followers to ones that have hundreds of followers. And being able to do that in scale requires that powerful search and discovery so we can surface these moments and then we can find and understand what about that content made that performance so special and how can we go rinse and repeat and go find more like that. So we can continue to kind of magnify our brands' voices and our brands' campaigns. None of this would be possible without aiWARE.

Fantastic. A question from Darren Aftahi asking how we're going to be compensated on MARVEL.ai? And what's the pricing? And how is it going to work from the ecosystem basis?

Yes. So as you saw when we went live today, we're kind of offering a very frictionless free trial for groups to participate, start playing around with the functionality and the different modalities, speech-to-speech and speech-to-text. So we're going to be pricing. It's going to be a combination of SaaS-based subscription-based pricing that's going to have X number of voices that you're going to have access to, X number of characters. We'll be providing more data on our rate card there. But we're really focused initially on the custom voice creation, right, for, I'll say, advanced speech-to-speech and text-to-speech functionality. And we're going to be working directly with our customers to really build those proprietary models. They'll all be ultimately deployed on aiWARE but again, our main focus is coming out with kind of the best-of-breed. And so -- and ultimately, those services, those managed services offerings for unique custom voice creation will include, again, both monthly subscription and SaaS fees but also, there'll be some managed service fees on top for the efforts to actually stand up and create the initial voices.

Great. Well, guys, I want to get us moving along. It's getting a little late in the morning. We have the financial section coming up with Mike Zemetra, and then we will have one more general Q&A, so we'll be able to hit the rest of the questions at the tail end of that. But let's move on and get the financial stuff out of the way, and then we'll come right back for another Q&A session to wrap things up. Mike?

Thanks, Brian. Before I actually get into the financial details, I would like to first explain that everything we do today is directly or indirectly supported by our aiWARE operating platform. Without the benefit of the aiWARE operating platform, there is no way we would be growing our SaaS software, advertising and licensing services at year-over-year growth rates in excess of 50% in Q1 2021. The aiWARE technology and platform is the underpinning operating platform around everything we do at Veritone. With that said, I thought it would be good to first discuss our 3 major sources of revenue in more detail: aiWARE SaaS software solutions, advertising and licensing. aiWARE SaaS software and solutions is the backbone of our technology, includes SaaS software and solutions where revenue is generated from existing verticals today including media and entertainment, GLC or government, legal and compliance and energy, sources of revenue or a mix of recurring SaaS, enterprise software and, to a lesser extent, from professional services. Revenue is recognized as the underlying software and services are delivered, which, in some cases, can be day 1 with enterprise software and/or over a contract term with recurring SaaS. Gross margins for SaaS were 72% in Q1 2021, up 55% year-over-year and are largely usage-driven. The improvement in year-over-year SaaS margins was driven by driving efficiencies in unit economics, coupled with incremental growth on new bookings, which is over 80% today. As we will discuss later, we expect our aiWARE SaaS revenue to grow greater than 65% year-over-year, with a long-term CAGR greater than 40%. Advertising is largely driven by Veritone One, which today generates over $300 million of gross billings per year. And on average, was up over 30% on a customer-by-customer basis year-over-year in Q1 2021. In addition, we recently launched our VeriAds network in the first half of 2020, which today now represents approximately 25% of our advertising revenue. Gross margins for advertising are greater than 90% for Veritone One and 25% to 30% for our VeriAds network. Revenue is typically recognized as advertising services are delivered. Overall revenue growth year-over-year was substantial in Q1 2021, with advertising growing over 70%, far exceeding the industry standard, which is above 15%. Our advertising services are mostly in the broadcast market, which include radio, podcasting and digital influencers where our audio-based aiWARE solutions dominate. Over the long term, we see large advertising opportunities across television and other IoT platforms as we take our aiWARE technology into newer markets. Lastly, content licensing. Today, we are powering the digital licensing of content libraries of some of the largest media companies in the world, including the likes of CBS. Today, we have over 3.95 petabytes of audio and video content digitally hosted and growing, which would not be possible without the pet power of our aiWARE platform. Revenue for content licensing is transactional and will vary depending on live sports and other events. Revenue for content licensing is recognized as the underlying services are delivered. Gross margins are in excess of 40% and can vary depending on the mix of revenue and live events. Growth of content licensing is largely a function of the size and type of content available and grew 18% year-over-year in Q1, largely due to more events we started to exit COVID. With the future advances in aiWARE and with the introduction of MARVEL.ai, we are looking to leverage our content library into new forms of content, news and entertainment. Today, the long-term CAGR for licensing is small. However, with the introduction of MARVEL.ai, the long-term CAGR could be approaching the AI market at plus 35%. Aside from the global digital ad spend market, which is in excess of $300 billion today, I wanted to spend some time providing an overview of the artificial intelligence market and even more time in the enormous and growing energy and government opportunities. The AI market is set to exceed $100 billion by 2025, with a CAGR over 35%. We think this market is much higher as it does not include the power of subvertical markets where AI will drive increased costs and revenue synergies over the next 4 to 5 years. Let's take energy, for example. Today, we know that over $26 billion in spinning reserves is lost each year as there are limited, efficient or systematic means to capture lost energy across the grid today. Moreover, we also know that billions of dollars are going to be invested over the next 10 years as more traditional power providers migrate to more green platforms such as solar, wind and battery. To that end, we have conservatively put our TAM in energy between $7.8 billion to $12.2 billion, with our near-term SAM at approximately $1.2 billion as we focus today more on utilities. Turning to government. According to Bloomberg, the government is set to spend $4 trillion this year. That spending is largely consumed by the Department of Defense, which tends to spend $750 billion, and the Department of Justice. Together, these 2 parts to the government, which are our target markets, represent nearly 1/4 of the entire U.S. government spend in 2021. With the mandate from President Biden, the U.S. government needs to do more with less. Veritone is in a unique position to immediately help the government solve its most challenging problems, all with AI. Over the past 6 months, we have seen parts of the DoD and NASIC endorse our aiWARE platform to solve incredibly challenging issues with existing satellite-based intelligence systems. In addition, we have seen the DOJ name Veritone in several task forces, including giving aiWARE an ATO, or an authorization to operate. In 2020, the DoD alone spent over $3 billion on AI projects and plans to double this next year. Our heavy channel partner strategy has provided us with immediate revenue opportunities where we have seen our sales pipeline more than double since 2020. Our fundamentals. At the midpoint of our financial guidance for fiscal 2021, and we are projected to end 2021 year with a CAGR of 54% revenue growth since 2017. Moreover, we have demonstrated solid sequential revenue growth over the trailing 4 to 5 quarters, including year-over-year revenue growth of 8% in Q2 2020, 23% in Q3 2020, 35% in Q4 2020 and over 50% in Q1 of 2021. That growth was fueled from our aiWARE platform, with aiWARE softwares and solutions growing over 50% year-over-year and advertising growing over 70% year-over-year in Q1. Our consolidated revenue growth is now projected to be over 40% year-over-year in fiscal 2021 using the midpoint of our guidance. This is up from 16% in 2020 as compared to 2019. Our bottom line results have seen massive improvements, including Q1 2021 non-GAAP net loss improvement of over 40% year-over-year. These improvements were driven by the expansion and growth in our services and gross margins, where we saw over 50% improvements in gross margin percentage for aiWARE SaaS and software solutions in Q1 2021 versus 2020. Driving this is responsible growth, the power of the aiWARE operating system and a very intense focus on customer retention and economics, which I will elaborate next. We are in the earlier stages of evaluating our customer metrics, but the trends we are seeing are so impressive that we wanted to provide you with an overview of some very attractive aiWARE SaaS and software solution customer results. First, you need to understand that we are an AI operating platform. As an operating platform, we are not throwing bodies at large projects that do not have recurring use of our software. This is a massive distinction versus some of our AI brethren, who will announce the large consulting projects. However, rarely are these consulting projects led with the likes of an aiWARE solution or recurring in nature. Given this, we wanted to talk about long-term value or LTV. The focus of our aiWARE operating system is to attract new and existing customers and land and expand in terms of products and usage across the platform. Using recent data in Q1 2021, our average revenue per SaaS customer is an impressive $115,000 when annualized in Q1. Our gross retention is outstanding at over 90%. We use gross retention to evaluate the impact of customer churn as we do not have a homogeneous revenue stream such as a subscription platform. Lastly, our net retention, which is a proxy for same-store sales growth year-over-year, was over 120%. These results put us in an elite category of SaaS providers. Using a gross margin of 72% and gross retention rate greater than 90%, our LTV is approximately $1 million, which is outstanding. It is important to note that our sales and marketing efforts around SaaS have largely been channel-driven, with only 20 or so salespeople supporting today. While we do not break out our CAC, you can infer that our LTV to CAC is also very impressive. Finally, we wanted to provide you with a long-term outlook on the business. First, the investment thesis is strong. As mentioned before, we're in an attractive $100 billion-plus AI market growing 35% through 2025. We've got a proven management team across the board, who you met on today's call. Our gross margins are over 70%, with 80% incremental gross margin on new customer bookings. We have no debt and over $128 million in cash today. Our growing SAM and market share within existing energy, government and M&E market is substantial. And our channel partner strategy is leading to doubling our revenue pipeline since as early as December 2020. Next is the outlook over the next 5 years through 2026. We expect to generate over $500 million in revenue by 2026, which is a 44% CAGR. As previously discussed, this growth will come from existing customers. Recall our net retention is over 120% today in new customers. Moreover, we expect to be a mix of organic and nonorganic growth where our revenue CAGR has historically exceeded 50% from 2017 through 2021 midpoint forecast. With our gross margin improvement driven by net incremental gross margin growth of 80% for new customers and operating expense growth limited in sales and marketing and R&D to 22% to 25% of revenue and G&A growth scaled down to less than 13% of revenue by 2026, we expect to achieve operating margins of 20% or close to $100 million by 2026. Under this model, we will reach breakeven by the second half of 2023. We're super excited to show this to you for the first time, and I know investors have asked for it for quite some time. And now we'd like to open up the line for Q&A.

Well, what a day. A lot of information, a lot of presentations. Obviously, a great finish with Mike with the long-term outlook there. I'm going to let the questions come in for a little bit, and I want to come back to some questions we didn't get to earlier on the energy front. Question from Mike Latimore asking, how much integration or interaction does our system have with the traditional management systems that are on the traditional grid? And how do we tie into those systems? Maybe chad?

From an installation standpoint, very, very little. Really, our software and solutions sit directly on the hardware providers themselves. So that's the inverters, the transformers, the PPEs, the batteries, et cetera. We do supply our own software interface that allows them to get a bird's eye view of how those optimization and prediction models are operating, again, both at the specific device level, but also aggregating up to the subsystem. But given the fact that we run on aiWARE, we easily can export that data directly into their DERM systems and SCADA systems to provide kind of a more holistic view for those operators. So yes, we do interface, but we're not required to actually do any of our deployments to install.

Great. And I guess kind of a follow-up from Darren Aftahi on that front is, within the energy space is the vision, the ultimate vision, to be having aiWARE built directly on to the OEM energy hardware and therefore, make it a long-term standard within the industry. And are there any obstacles to having that come to fruition?

Yes. So that's our Phase II strategy. Right now, the energy grid, it does not operate like a plug-and-play network. We developed a new inverter, someone does, and they want to plug it into a network or I come up with my own IPP, independent power provider, that I want to install and integrate into an existing grid, there has to be a substantial amount of both software and hardware handshaking to happen before that works. And that's just not an untenable situation given the demand for independent power. And so what we're seeing now that we're installing with major grid operators is they are forcing their providers of hardware, whether they be the solar rays, the battery manufacturers, the inverters to adopt our controllers and integrate them directly into their hardware. So we think this is an ongoing trend, and our Phase 2 would be to work directly with those hardware providers so that essentially, it becomes the plug-and-play standard controllers and prediction models. So that no matter what grid they plug into, they're assured that their technology will work with that operator. So that's absolutely where we're headed, and we're going to be pricing our services at the OEM level, almost like Dolby does, to become just that standard to ensure plug-and-play for the future of the green energy market.

That's great. Mike, this one's for you, it's coming from Mike Latimore, talking about the sales pipeline. We've talked about that doubling since December. Can you put that in perspective, how much did that pipeline grow over the course of 2020 relative to obviously doubling here in the near term? And then within that, he has a follow-on question with regards to the MARVEL.ai revenues. Will those be recognized in the SaaS portion of our business? Or where will we see those?

Yes, terrific. Thanks. On the first question, the sales pipeline, when I talk about it doubling, I'm specifically referring to GLC. And so what we've seen in GLC over the past 3 months, including task orders coming to us from the DOJ, has really opened up massive opportunities across our channel pipeline. And so that's where we're really seeing the inertia. And just anecdotally, Q1, the growth in GLC year-over-year was over 150%. And if you look at Q1's revenue in GLC in comparison to the prior year, I think it was over 60% of the revenue generated all for the entire of 2020. So we're certainly seeing healthy there. And then on the energy front, there's no shortage of demand. So we're seeing, in terms of pipeline, we're not going to quantify it on today's call, but it is more than doubling where we were just 3 or 4 months ago. In terms of aiWARE and the interplay of MARVEL.ai, I mean, listen, we're certainly in the nascent stage. But this is going to be SaaS. It's going to be a combination of SaaS and other things. The power of MARVEL.ai isn't going to just impact one line item in the long term, it's going to impact all of our line items. So at least in the early stages, it will be directly in the aiWARE SaaS category, but it opens up lots of opportunities for additional revenue, incremental growth across the other categories as well.

Ryan, do you maybe want to follow-up on that in terms of how MARVEL is going to move into the other categories as well? Just flush that out with your vision?

The other categories, the other BUs, is that what you're referring to?

Yes.

Or in different revenue lines. Okay. Well, I mean, right now, we're just kind of solely focused on primary SaaS, right, as licensing the technology and incurring it. But there's a lot of opportunity for us to leverage this, what we're doing in-licensing. Obviously, in Veritone One, we spend a lot of money. Our clients spend a lot of money, time and effort to produce advertising copy, content, branded content, so we -- so again, our main thrust and focus right now is on, I would say, SaaS-based, technology-based revenues. But again, we already have demand for this to be extended and made available for licensing, augmentation and activation as well as advertising capabilities. So we're excited. And then GLC, that would still fall under SaaS-based revenues, but providing the technology to empower the local municipalities in local jurisdictions with the ability to more efficiently communicate, right, to their populace is something that we're excited to immediately push through our channel, which Jon Gacek and team have already established for us in local government. So we're -- so we -- again, we think this will have broad appeal and broad applications across the enterprise.

Fantastic. A question for you, Mike, with regards to the LTV conversation. Nick Mattiacci is asking with the retention rates given on that slide, are those actual results? Or are those just for calculation purposes? The way you spoke in your script, it sounded like it was actual, but the small print at the bottom says something slightly different. Can you maybe clarify?

Yes. So I just want to caveat that it's preliminary data, but it's based upon actual Q1. So this is all actual results in real time.

Perfect. Great. And then looking at the clock here, I think we got time for one last one. And Chad, I'm just going to warn you, this is coming from Pat Walravens. This is his typical future question. What are the top 3 things you want people to walk away from today's presentation with?

I think it's pretty straightforward. If you listen to the whole thing, it's aiWARE, it's aiWARE, it's aiWARE. At the end of the day, when we sign a customer, it doesn't matter what their pain point is that we're trying to solve, whether it's in the government legal and compliance and redaction or media entertainment with advertising or even MARVEL.ai. When you sit down as an end user and you start plugging that into your existing workflows and systems, you start moving that human from being in the loop to on the loop and eventually out of the loop. Inevitably, you are going to want, as an organization, as a customer, to consume more AI in different workflows, moving different humans from in to on to out of the loop. And aiWARE, just like Windows, just like MacOS allows that user to not only just reach up and go to a marketplace and pick a different application to start to extend the power of AI in their organization. But with our developer suite and our partners and our system integrators, they can now start to develop their own solutions and their own applications and their own integrations to really start to drive that value home. So at the core of our business is aiWARE no matter what it is and what industry we're focused on. We're excited to look at new markets. That, again, when you think about acquisitions or partnerships with different groups, without an operating system, you're basically cobbling together bespoke one-off solutions that either don't scale or won't interface well with other tools. With aiWARE, when we start thinking about integrations and partnerships, we have a playbook that we've demonstrated now more than 2 dozen times to be able to uplift and integrate applications onto an aiWARE stack that makes it a native platform. So again, I'm excited about the future. We've made a tremendous investment over the last 7 years in this operating system. And I think we're just starting to reap some of the benefits of that, that you're seeing from the financial results that Mike discussed.

Fantastic. Guys, it's been a heck of a day. We started out the day with over 200 people. We finished with just about 200 people. A ton of information, a great outlook. Thanks a ton to all of those people behind the scenes in the marketing department, the IR department, the finance department, fantastic work. And to all the people I didn't get to your questions on, please just shoot me an e-mail. Shoot it to [email protected], and I'll do my best to respond directly. Thanks, everyone.

Bye guys. Thank you.

Bye. Thank you.