Twilio Inc. (TWLO) Earnings Call Transcript & Summary

Thanks, everybody, for being here. I'm going to read a real quick disclosure, and then we'll head into the session. Please note for all important disclosures, including personal holding disclosures and Morgan Stanley disclosures appear on the Morgan Stanley public website at morganstanley.com/resourcedisclosures or at the research desk. I am Meta Marshall, I cover the communication software space here at Morgan Stanley. We have Jeff Lawson, Founder and CEO of Twilio, with us here today.

So I'm reordering the question list just a little bit to maybe kick off with a coronavirus question and just any impact that you're seeing or how you're looking at the potential impact to your business.

Absolutely. And first of all, thank you for having me here today. I want to begin by saying I feel fine and I went to -- but I'm saying this for a reason. I went to the trainer this morning for a gym session, and he asked me to help him rewire his Sonos. I'm serious, like I'm the IT guy for my trainer. And I got on this dusty, like, back of a gym and I literally like I've got -- I'm -- runny nose and sneezing, I swear, I am not sick. But it appeared like right after all this dust was in my face. So that is why I have a Kleenex in my hand. That said, let me talk about coronavirus and Twilio. So first of all, we've obviously, as everybody has, we've been very focused on keeping up with the news and understanding what the best advices that everybody has. Our first priority is on the safety and well-being of our employees. Obviously, everybody is very concerned. And there's also an aspect of psychological welfare, too. I mean people don't know if they should be concerned. Whether this is just essentially the flu with a lot more media and a name on it, or whether it is something they should be concerned with. So we've been monitoring that and making sure our employees feel safe physically as well as psychologically. Then as far as the business goes, I think that, obviously, we don't know exactly if there will be an impact on the business. We can imagine scenarios that -- where there is an impact because if travel or hospitality industries are impacted, while we have those customers, and if there's less usage of their products, then that could mean less usage of our product. We can also imagine some upside, wherein, for example, in the on-demand delivery of food, for example, people are staying in, instead of going out, that can mean more usage of certain use cases like that. And obviously, we're in the electronic communications business. So if there are scenarios where people are using more electronic communications to communicate as opposed to in-person meetings, that could be an upside. The net result of all this is, is obviously, we don't know. We're not changing anything about how we look at the business right now, but we are continuing to monitor and work with our customers on making sure their businesses are well and seeing if there's any additional ways that we can help them during this period of time.

Great. And I think I'm really excited to have you here today. Because I think when I talk to investors about Twilio, it tends to be one that has a lot of misconceptions about it. And one of them, I think, is that the CPaaS market is simply 2-factor authentication, it's delivery alert. What do you think investors miss most about the breadth of the services that CPaaS is? And what is it used for today?

Yes. The amazing thing about the market that we're in -- so the critical letter in there is the P, platform, meaning that you can build just about anything with a communications platform as a service. And so while, obviously, there's the high visibility of certain use cases like 2-factor authentication or one-time passwords, it's the full breadth of communications. Every point in a customer life cycle, where a company engages with you. And that could be during the marketing, trying to reach you, it could be during the sales process of trying to close you as a customer. That could be in the delivery of the product itself, so when you're using the product. Or it can be in a service and support arena when you need help or -- and even someone's rolling a truck to you for field service and logistics. These are all the touch points that a company has with you, its customer, and every one of those touch points can be made better with better communication. And that's really the notion of customer engagement. And so we see companies across the spectrum, implementing Twilio to build better relationships with their customers because of the right moment, at the right time during your engagement, they're being proactive in communicating with you. In fact, I got a text yesterday saying I made a dinner reservation at a restaurant and I got the confirmation text, and they said, "We'll text you again, the day I have to remind you, and we'll let you know when your table is actually ready." Like, very useful. But the interesting thing about it is that is the notion of like alerts and things like that, those are oftentimes the first interaction we have with the company. Why? Because the amount of work a company has to do to implement it is actually very small. And the benefit their customers get is actually really huge. So imagine all these sort of opaque processes where you don't really know what's going on inside of a company. Like you don't know, did they actually get my dinner reservation or is my table ready, and just to use that example. But now I'm kept in the loop. And I get a proactive alert. And so for companies, first of all, it reduces the workload on the contact center. People contacting them to say, "Hey, did this thing happen or not?" Well, I'm practically keep you in the loop by letting you know. But then the interesting thing of like that is a common first use case. But that leads to the great question of well, what happens when a customer replies to that message? And most companies haven't thought that through yet. As you can tell, we get texts all the -- it was not long ago. 4 or 5 years ago, maybe it was novel, when you got a text alert from a company saying, "Hey, your package is shipped," or, "Your dinner table is ready," or, "Your dry cleaning is ready," or whatever it is. "Your flight is delayed," or yes, well, your flight is on-time, actually. But that was novel 4 or 5 years ago. Now we're starting to come to expect it. We're starting to come to expect it from just about every company that, oh, yes, of course. It's kind of weird actually if they don't keep you in the loop when things are changing. But so far, you don't actually have the experience very commonly where you can reply to that text. Think about it, take one of those texts. You currently have one in your phone from the last 24 hours. Reply to it, and see what happens. Sometimes you get like some standard response back, like, not monitoring this, text and data rates may apply. Sometimes you get an answer that says, we're not monitoring this, call us on our counter -- they give you the phone number, "Call us here," or, "E-mail us here," or, "Go to this website." You're like, "Well, I'm replying to the text you sent me. This is the convenient method. I know you have a website, and I could have gone to. I don't want to do that, I just want to reply to you." But the most common thing that happens when you reply to one of those texts is nothing. Just goes into the ether and you sit there wondering is someone like sitting there in a contact center, thinking about how to reply to me. Is there some bot that's kind of working right now trying to -- and like the answer is usually like, "No, they're even wired off." No one is even listening when those messages come in. No one and nothing. And so that's an amazing opportunity that's on the horizon. And so I think that when you think about the fullness of all the use cases that companies have for how they communicate with their customers, a low-hanging fruit one is these alerts and notifications. That's great. But then it opens the door to ask but what other parts of the business actually depend upon you having a great dialogue with your customer. And I think about the marketing use cases that are especially prevalent now that SendGrid is a part of Twilio. And for those of you who don't know, Twilio's platform, our very first product was voice and then we had an SMS, and then we had a chat and video and WhatsApp and Facebook Messenger. But e-mail had never been something that we had touched because there was another great company actually doing that as a platform as well, they're called SendGrid. And they're the leading developer API to send and receive e-mail in applications. And so last year, we acquired them and brought them under Twilio. So now we have the leading API for e-mail as well, also as a part of Twilio's product portfolio. And as a result of that, we're seeing more and more of the marketing workloads because many companies start their digital communications with e-mail. And especially in marketing, that's a pretty common thing. You often see the sign-up for our newsletter as a prominent thing. And so many companies marketing strategies really start with how they engage to an e-mail. And so with SendGrid, we're now starting to get a great window into the e-mail of the marketing workloads. We already had a lot of companies doing sales with Twilio. For example, the call tracking use case. When you see a phone number on a billboard or a website or anywhere, and that phone number is actually a trackable phone number. So that when you call into that phone number, they can see where the communication came from. They can see that their ad buy on this channel is working because it drove inbound interest. So that's a use case. That was one that we started with the early days of Twilio. And then what happens is as you take them on phone call, you are forwarded to a contact center. Then we started talking to customers about, "Oh, okay, well, what contact center are you using?" And they would tell us, "Well, actually, let's talk about that as well." And so you can see how you start with these early use cases that are easy to implement. And then that opens the door to be able to do more, and more, and more with customers. Like that's a similar precession that led us to develop Flex, our contact center platform that we built. So anyway, it's across the spectrum. And it's sort of interesting when you look at companies. Every company starts in one spot. Because there's some problem they're working to solve and a developer brings Twilio in and says, "Hey, I know how to solve it." And then with our sales team, we go in, we make them successful for that first use case. But because we're a platform, you can build just about anything. And so we go in and then we figure out, great, where is the next use case and the next one, and the next one. And that's the power of a platform business model.

Got it. So maybe -- super helpful. And then perhaps leads us into the second question which is just a second misconception people have is that CPaaS will inherently be a low-margin space, particularly given some of the other vendors in the space. How do you think that that misses the opportunity you have with an organization to offer these higher-margin services, some of which you just kind of described?

Yes. So communications, first of all, is not a new industry. It's been around for 100 years, right? And so there's always been folks who are like carriers, right, who essentially transit bits and wires. And that has been -- and that's nothing new. Obviously, there's AT&T, there's CLECs, there's a bunch of different types of providers there. And I think the difference with Twilio is that we started the company in a very different mindset than like a carrier because myself, Evan, John, the 3 cofounders of the company, we were all software developers. And our fundamental question was how can we, as software developers, build communications into the apps we're building. Because communications was this foreign thing. I'm a serial entrepreneur. I have started multiple companies before Twilio. And in every single one of my companies, at various points in building up that customer journey, we would say, oh, wouldn't it be great if we could communicate with our customers like this or like that. And every time we have these ideas, we would say, "Yes, that would be really neat, but I'm a software developer. I don't know first thing about communications." And so we didn't build those things. And so we built Twilio to solve the problem of letting software developers actually go build communications into their apps. And so we are a software company founded by 3 software developers who actually didn't know very much about communications when we started the company, and built out all these APIs and this whole software layer that does all the intelligence of what does a developer in a company who's building communications and software need to just to, A, build it. So build out the use cases, all the APIs and all those features and functionality as APIs. And then because it's software, you really have to think about what is world-class software look like. It's a distributed system across -- I think we run in 28 data centers now around the world, and you've got a high uptime requirement. You have to be always available. You've got security requirements that, obviously, companies have. And like in things like even today, we just announced that we are now signing BaaS to support health care workloads on Twilio. And so these are really big, expensive software engineering feats to do these things well in a global scale, that is about a software exercise. It's not about copper wires and fiber optics. And so that's the perspective that we've always brought to Twilio. I think close to half of our company is our software engineering, our R&D organization. And that's the building the products that we build, operating the products at global scale with high availability. Now let's contrast that with a carrier. Right, if you're a carrier, we've seen some folks with carrier backgrounds enter the space. But really, what they're doing is saying, "Hey, I got some pipes, how can I sell them to a different audience?" And so what most of them have done is built a relatively thin veneer of software on top of those pipes and say, "Hey, we're CPaaS," and say, "Hey, you can come and use it." But the problem is when you're a carrier who generally has lower gross margins, and so you don't have a lot of gross profit to use to build a world-class software engineering organization and invest in all that software, whether it's features, functionality, reliability, HIPAA, ISO 27001, software things, well, you do the minimum required. And then because you don't have the features and functionality or reliability or security, you say, "Okay, well, it's cheap." That's our benefit. And I'm sure you're going to get some customers, absolutely. But when it's cheap, you don't throw off the gross profit to actually invest in the things customers really care about in the software. And so that kind of propagates itself. You can't build all the things that customers need. And so you continue to focus on price, whereas we, because we started as a software company, always internalize the value of what software platforms bring to the world. And what software buyers are looking for when they buy software, that's commanded a healthier price that allows us to reinvest the gross profit back into building more software for our customers. And that also keeps propagating itself. So our approach has always been to come at this market as a software company and to serve other companies who need software like features, functionality, reliability, security, et cetera. And I think that approach has served us very well in this market.

Got it. One of the most interesting move, I think, you guys have made is bringing on board members and execs who don't necessarily have a Silicon Valley like insular or pedigree. And so you have access that organizations with larger digital transformation needs. How did you think about kind of that as an angle of adding value to your company?

Yes. I just -- I think it takes a village, and Twilio is a pretty unique company, like we don't actually just pattern match exactly to one other model that's out there in the world for better or worse. And I obviously think it's for better. But that has meant that as I've gone around building the team, whether it's the executive team or the Board, I want a diverse set of viewpoints from different types of businesses and different types of industries to help inform how we build this company. And so I'll give you an example. On the exec team, I'll talk about a few different folks. We have Khozema, our CFO. And Khozema came to us from GE, which is a very operationally focused company. And as we grow and we just crossed $1 billion a year, last year was our $1 billion -- not run rate, $1 billion revenue milestone year. I was -- the first thing that comes to mind, as I say, "We crossed it for the first time," but nobody ever wants to hear it. Did you cross it again? Like that would obviously be bad. So I thought, like, not say that. But -- so we crossed the $1 billion. That means it's a different level of scaling and operational maturity you have to build in the company to scale. And that's why Khozema is a great CFO for the company. On the go-to-market side, we've George Hu, an executive who came to us from Salesforce, where he was the COO at Salesforce. And he's brought a lot of the go-to-market DNA that has helped the leading SaaS company in the world grow to the size that it's at. And so that's been -- George has been a fantastic influence on us, even though we're a platform. Now let's talk about the platform. Chee Chew is our Head of Product, and he runs all of the R&D organization. And Chee comes from a strong technology background. And immediately before this, he was at Amazon, where he led customer engagement for all of Amazon retail. Before that, he was at Google, where he actually built the communications products at Google like Hangouts. And then before that, he was at Microsoft. And so here's someone who has seen and built world-class engineering organizations at the iconic technology companies of our time. And so you can see how this team all bring different perspectives to the table as we're building the company. On the Board, I think it's interesting. We've got a number of unique voices. One that's interesting, and so a lot of folks have asked me questions about, last year, we added Jeff Immelt to our Board. And obviously, a little bit of a controversial character. But from my perspective, he is a world-class executive, of which there are very few human beings on this planet who have operated at the scale and complexity than what Jeff Immelt has done. And you can argue whether or like, whose fault it is, for the outcome, but regardless -- first of all, I'll just say it this way, if GE was continuing to be a skyrocket, I wouldn't have him on our Board. There'll be one less brain to be able to tap into building this company. But the fact that we have him on our Board, he has been fantastic in helping us see through global expansion issues like how do we think about running a truly global business. There are very few global businesses, the scale and the complexity of GE. And then as far as enterprise, there's very few people in the world who actually know all of the big the CEOs, the executives that nearly every company out there like Jeff does. And Jeff, I don't know if you know this, he was a salesman at GE. So he's got a fantastic sales DNA. So helping us think through as we think about the next 10 years of Twilio and what are we going to do to crack the enterprise customer because our goal is really to be the wall-to-wall, end-to-end customer engagement platform for enterprises and customers, he's helping us think through the execution of how we go do that. And so I just think it takes a full team with a breadth of experiences and the breadth of different industries that actually help us build this unique company and I'm really appreciative of the people that I get to work with every day.

And so maybe you've -- building from that, you've had a developer-led model, which gives you great inroads into kind of the tech innovator company is in this space. But as you're talking about your vision for the next 10 years, will it always be a developer-led model? How should we think about the timeline of bringing some of these larger organizations on as they look at their digital transformation effort?

Well, I think we're already seeing it happen. So it's not a matter of like, okay, well, we can't do it today, and then 1 day in the future, there's -- we can do it, right? Like all things, this happens over time. But I actually think our product is much easier one to bring into large sophisticated enterprises than a lot of companies. If you think about -- think about like an HR software. A company has one of those, or a financial software, a company has one of those. And so it's a very big decision. It's a very long time frame, not just to buy it but to implement it. And it's fiercely competitive because there's only one that the company has. So every vendor in the space is going to be in there like tooth and nail trying to get that business. Whereas communications, there are workloads all around companies. And so many greenfield opportunities. So when I think about the ways in which we've been working with a lot of our enterprise customers, it's been -- like there's always a workload. There's always a customer experience point that can be improved. And there's someone at that company is thinking about how can we improve it. And they bring in Twilio and developers start the process. And they bring in Twilio and they often build the prototype on their own, and that gets the ball rolling. Now if it was like a tech company or a start-up or something like that, maybe that was just the end of it. They would just deploy and run on a credit card for a while until we call them. But in enterprise, sure, you're going to run a sales cycle. And typically, there's going to be a decision-maker. But it's a lower down decision-maker on day 1. It's not the CIO who's deciding to put in alerts for a certain use case. That's a great first foot in the door because it is a low decision maker. It's low risk, and it's fast. The developer can usually build those use cases in days or weeks, not months and years. And so it's a fantastic way to get in there. And I love even with Flex, our contact center product, we've had some great wins. If you look at the last few quarters of our earnings call, some of the customers we've been announcing, Allianz in Q3. So Allianz is using Flex to engage with their customers over digital in their direct-to-consumer business. And we had Compucom that's a division of Office Depot, who is using Twilio for all of their -- for their -- essentially their pro serve, their like full-service business they have with serving companies. That's what CompuCom does with Flex. And then even companies like Southwest Airlines pulling on the Flex. And so it's been a fantastic run of taking not just the building blocks. Like, some of those companies have been customers of ours for a long time. Another adding on in my opinion, they're getting more strategic in their use of Twilio because we're going from, again, that footnote or use case of sending you an alert to actually powering the contact center. And then that opens up more conversations about, okay, well, where else can we take this? Oftentimes, the contact center begins with one of their contact centers. We often have many contact centers to get your foot in the door with one use case, then you expand into one of their contact centers, and then you can see the, okay, how do we take over the rest of those contact centers. And so it is a land and expand, for sure, and we are seeing the land. We're seeing it happen every day. And then we continue to see the expansion happening. And so I think this is happening very nicely. One of the things that we've done that we innovated on last year that has worked really well, and we're expanding it this year, is our enterprise hackathons, where we actually bring Twilio into a company. And we hold a 1 or a 2-day hackathon with their engineers. And we sort of said, like we tee it up as, "Hey, business executives, why don't you throw out some of the biggest problems that you're looking to solve for the business?" And then what your developers work for a day or 2, showing you building some proofs of concepts and some prototypes, and then come back at the end. Let them show you what they were able to accomplish in just a day or 2. And it is amazing. First of all, it's just amazing to see the developers and what they can accomplish. The executives are usually so excited because what -- I mean every executive wants more progress faster. And when you actually show them what can happen in a day, it paints this picture of what's possible for them, not that you're going to solve the problem in a day, but it shows you, oh my -- look at this progress we're able to make. Look at what our product could look like. Let's let this keep going. And so those have empowered the developers, empowered the executives and for us, built a lot of sales pipeline. And so those have been really successful programs and now we're looking to scale that up in 2020.

Got it. And so maybe building on the Flex conversation that we just had. Some investors, I think, have gone through the high cycle of their own of saying that, in their mind, Flex should have ramped faster, I think that misses, generally, that sales cycles can be very long. Has Flex business developed differently than you've thought, both in terms of the pain points that customers are trying to address or what customers have been using it for?

I think Flex is progressing very nicely. The hypothesis that we started Flex with has been proven true, in my opinion, which is that companies historically have had to choose between their old on-prem contact centers, and this is a market that is still 85% on-prem, and companies have had to choose between sticking with their old on-prem software because on-prem allowed them to vary -- to customize the contact centers substantially. So they could take a contact center with enough professional services come in and really bang that thing into submission to make it do exactly the workflows they needed to do, which makes a lot of sense, right? When you're a company with 10,000 contact center workers, 20,000 contact center workers, every optimization you make to those workflows to make the agents more productive, to make the calls with customers shorter, you multiply that by 20,000 people every day, it's a huge ROI. And so companies have spent the last 10 years dialing in all these workflows to make these contact centers extraordinarily efficient. And that's what they got out of on-prem. The problem with on-prem is that every customer is running their own bespoke version of the contact center. Partially because they customize it so much, but then you also throw in on-prem software and upgrade matrices and what you want to -- you want to upgrade this, like one of the vendors is like, "Oh, we have a security update, you have to apply it to this part of the stack." And then you do the math, and you realize the whole stack has to get upgraded in order for them to interrupt with each other correctly. And they'll spend -- regularly, different customers, they spend like 9 months just trying to do an upgrade. There's no innovation in that. It's just keeping the thing running and secure, and that's the mess that they're dealing with on-prem. Yet, the cloud is designed to solve this problem, right? Here we are 20 years in the whole cloud thing, and 85% of the contact center market is still stuck on-prem. Why is that? Well, the answer that we heard from customers was that nothing in the cloud allowed you to do all those customizations. Because the very thing that makes the cloud great, multitenancy, it's a single code base, every customer shares the code base, meant that you couldn't actually go in there and really rewire how the thing works in order to customize those workflows. And so companies told us that they were stuck. They were stuck between customizability on one side and reliability on the other side. And guess what? They happen to want both. And so we answered that with Flex. And I think that hypothesis has now been proven up by the amount of customer demand that we've seen for Flex. Now there was a bit of a hype cycle, I think, especially in the investor world, and it's obviously flattering, this, launch of product and have everyone get so excited about it, more excited about the customers than necessarily the investors because customers are what lead to success for everybody. But I think the reality is that contact centers, they do take a while to play out. When we look at the revenue from Flex, it's not something we've disclosed, but when we look at it internally, I'm very happy with the speed at which this product is growing. And I think as an independent company, everyone will be blown away by the speed at which it is getting traction with customers. But I will say, it's still essentially a version 1 product. We announced it about 2 years ago, 2 years ago, if my math's right on that. Yes, 2018. And so we announced it, we GA-ed it a little over 1.5 years ago. It's actually said it's for sale. And so with that in mind, I think it's doing incredibly well as a product and it's got a really great customer base. But I'm not sure what investors' expectations were, but they may have been a little heightened given a V1 product just coming into market does take time. And the contact center, especially for larger ones, takes time for customers to decide and implement because you have a retraining to do at the workforce to use at that contact center. You obviously have the implementation time line, the decision time line. And so those are a little bit longer than some of the alerts and notifications use cases, but that's to be expected. That's built into our models and so we think Flex is progressing incredibly well for where it's at in the life cycle.

Got it. And then maybe just final misconception investors have...

A lot of misconceptions.

I know I'm trying to clear it all up because this is -- I think it's a great story. It's that customers think or investors kind of think that RCS or WhatsApp for Business, create kind of headwinds versus opportunities. So can you help investors contextualize how the opportunity is different in multimodal, like, communication and how it could be a bigger opportunity overall?

Absolutely. I mean it's no surprise, right, that we as consumers are using myriad ways to communicate with each other. We're using messaging, we're using -- but to do that, we're using iMessage, we're using WhatsApp, we're using Facebook Messenger. We're using all these different platforms. Then you have the geographic aspects of it. In Japan, they use LINE, in China, they use WeChat. And different parts of the world are using different apps to communicate. And so most of us on our phones have, I think, on average, 5 or 6 different messaging apps that we use for different purposes. And that continues to get more and more and more complex, actually every day. And so companies are trying to figure out how do I talk to my customers? How do I use all this new technology to engage with my customers? And the answer is you really have to listen to your customer. If your customer wants to engage with you on WhatsApp, you should do that, if they want text messaging, you should do that, if they want voice, you should do that, if they want chat, you should do that. I mean it really comes down to which channels are customers going to respond to? The best, that's the correct channel to use. And that's not universal. That is customer by customer. It's an individual decision that we all make. Some people love talking to brands in WhatsApp or Facebook. Some people hate it. They say, "Stay out of it. This is for my friends and family." And so it really is a personal decision. And so what you -- what the best companies are doing is building in a way where they can support all the channels that their customers may want and letting customers self-select into the channels that actually make the most sense for them. I thought it was interesting, one of our early customers on our WhatsApp API, so we have the Twilio API for WhatsApp. And one of our early customers was Deliveroo, one of the food delivery service in the U.K. and many different countries that are headquartered in the U.K. and when they -- they had offered e-mail, they had offered SMS. But when they offered WhatsApp, they saw uptake of that feature shoot up. I think it was 6x the uptake of any other channel that they had offered historically for actually people choosing to get updates on their orders and things like that. And I thought that was fantastic. So it just goes to show you that the more channels you can offer to that audience -- that was in the U.K., but in other countries, it was different. The more channels that you can offer, the better chance of a customer actually opting in and saying, "Yes, you know what? I actually want to engage with you because you're actually doing it in a way that I like." And that's a huge opportunity for Twilio because instead of having to invest independently in every one of those different channels and every one of those road maps -- and by the way, the company's Facebook or WhatsApp or Google with RCS or all these -- they're changing these all the time. If you actually were to subscribe to the changelog of some of these APIs or in the systems from these companies, you would see that, like on a weekly basis, you'd have to keep making changes to implementation just to keep things running. Well, what's great about a platform like Twilio is you build it once, and it works on the channels that you think you need today. And by the way, when a new channel comes along, can work that, too. So it turns out if your customers really want TikTok, fine. Like, Twilio will be the place that you'll be able to do that. That is not a product promise today. I'm just making the point that like by future-proofing your investment, you build on a platform like Twilio, and now as the world changes, and it's going to keep changing, and it's going to change even faster. Like I had a customer asked me once like, "What should my HomePod strategy be?" You're like, yes, like companies are like, "Do I need to invest in HomePod? Is that the next big thing?" And my answer, honestly, is I don't know. But if you build on top of Twilio, if it turns out that HomePod is important, we will support it, and you don't have to do anything because you are already built against a platform that is omnichannel by design. And that means that you get to build once and then be future-proofed in that investment so you don't have to keep making bets on your strategy about which channels your customers may or may not want. And people have been burdened so many times. I remember chatting with a customer back in -- that was about 2011 or so shortly after we launched our SMS product. And the customer was hesitant to use -- to adopt SMS, and I asked why. Said, "Well, look, we're 2 years into our mobile app strategy. And we're about to launch mobile apps. And once we launch our mobile apps, we'll just be able to send push nonfiction for our customers. And that's why we don't need SMS." And I said, "Okay, well, let me know how it goes." And about a year later, I talked to the same customer. They have launched their iOS app and their Android app that they've spent years building, and he said, "Well, after that, we're a year in to having launch them, and I've got about 20% of my customers have downloaded the app. And of those, about 20% have actually said, 'yes, you can send me push notifications.' So I've got a solution that allows me to communicate with 4% of my customers. Now can we talk about SMS because I need to reach other 96% of my customers?" So it's like they had made this huge technology investment, not a bad one, I mean, obviously, mobile apps are important. But they have made this huge technology investment and got a fraction of their customers to opt into it. And so you keep going back to great, how do I make investments that are going to reach the maximum number of customers and not to have keep reguessing as those consumer behaviors change. And Twilio is that platform.

So maybe last question from me. AI is a great opportunity. I think mostly it gets brought up in contact center. Where do you see Twilio being able to participate either for what you can bring as part of Flex or what you can bring as kind of part of Twilio altogether to the communications world?

Yes. The biggest trend that we see, it's showing up largely in the contact center, but in general, in voice and even use cases like video is how do I take communications and turn them into data. But how do I take -- it's easy to imagine like chat sessions, like if you chat with your cable company or whatever, you can imagine them looking at the chat transcripts and like data mining it to understand what are people asking my agents and how good are my agents at answering it. And even that actually is still in its infancy, but you can use your imagination, say, yes, that's pretty easy to do. But then you take a voice phone call and you're like, "What do you think they know about the things that our agents are saying?" Or out in the field, if someone's doing a delivery and they call you, what do you think the company knows about the things that are being said between their employees or their contractors and their customers. And they know almost nothing. And so the biggest trend that we're seeing is how do you take real-time communication, like voice or video and things like that, and how do you turn them into data so that then, you can act on that data? So part of the data is just mining for who are sales reps or support reps, who are not performing well and how can we coach them. How can we actually coach them not after the fact, but also in real time. So can we transcribe these calls in real-time to know what's being said and then provide coaching to the agent? Can we actually like, when they have got their script of things they're supposed to say, they're supposed to greet you a certain way, they're supposed to sign off in a certain way. You've all had the experience where you are [ bucking ] up and they say, "Oh, wait a minute. I want to let you know that we've got a survey." It's going to come out -- like something you're supposed to close with. They have to remember to do it. And so you can actually provide that coaching real-time because they know, the AI can know, whether they've said those things or not said those things. And then you can actually start getting into dynamic coaching based on sentiment and tone and things like that. So if a customer is changing their tone, well, the agent can actually change their approach to the call. And the AI can coach them on that. And so you see all these amazing ways in which AI is essentially helping the human beings. And then you also see the ways in which AI can actually help the humans be more human. And I'd say that in the sense of you spend a lot of time on calls with people just exchanging information. What's your account number, what's your mom's dog's maiden name, whatever. And it's like, can I just tell that to a bot ahead of time? So by the time when the human picks up the phone, they're able to greet me by name and know why I'm calling. And that's really the promise. Let the bot do the bot things, and let the humans do the human things. And right now, we've got humans, like -- everyone have it backwards in some ways. You've got humans doing bot work, asking you stupid questions. And you got the bots trying to be humans, and that's not working particularly well either. And so if everyone sorts these into the right buckets, I actually think we'll make our human beings much more effective, much more productive and then we'll also have a great use for the bots. And I think the bots will get better, and they will start being able to do the human things better. But that's not where I would have started. And that's kind of where the industry did start in some ways, when I was a little bit of a false start there. I think I'm the bot hype that you saw in 2015 and 2016.

Got it. Well, Jeff, I want to say thanks for being here today and appreciate it. And we'll now take a break.

Great. Thank you very much. Thank you, everybody.