Salesforce, Inc. (CRM) Earnings Call Transcript & Summary

Everyone, welcome to today's session, 3 ways AI will scale your contact center, and thank you all so much for joining us today. My name is Ariana, and I'm on the marketing team here at Salesforce. And before we begin, I'd like to cover a few quick notes with you about our webinar platform. Today's webinar will be available on demand after we wrap up, and it will be accessible through the URL that you are on right now. Please note the slides will advance automatically throughout the presentation. And you can enlarge them by clicking the enlarge slides button located in the top right-hand corner of your slides widget. Should you need technical assistance, click on the help widget located on the bottom left corner of your console. We've also added a few additional resources which are available through the resources window to the right of the slide. There, you can find some additional related content. And lastly, we encourage you to submit your questions at any time throughout the presentation today using the ask a question widget at the bottom of your console. We'll do our best to answer as many questions as we can at the end of the presentation. And with those logistics out of the way, I am turning things over to Amanda to get us started.

Great. Thank you so much, Ariana, for your help in giving everyone logistics and details. My name is Amanda West. We have an all-star lineup for you today. I am part of our Product Marketing team and I focused on the Contact Center. And with me today is Oana Lungu Polanco, a Senior Director of Product Management. She's going to walk you through the 3 ways to get started in the 3 ways to scale AI in the Contact Center to deliver intelligent customer experiences. Next, Tamer Farag will demo how AI will make life a lot easier for not only your customers but also your agents and your supervisor. Last but not least is Robin Gareiss, CEO of Metrigy and also a principal analyst there. Robin will walk you through how AI helps your agents handle more chats and also drive revenue. So without further ado, let's get [Audio Gap] AI has dominated the tech headline and is arguably one of the coolest and most revolutionary technologies of our time. While generative AI might seem like a buzzword, there's actually a widespread consensus that we're on the precipice of a monumental change, even more so than autonomous vehicles or NFTs. Generative AI is unleashing a magnitude of far-reaching opportunities and also risks beyond what we can even begin to imagine. This powerful technology is creating exciting new opportunities literally everywhere. And what we are seeing in every company is really wanting to use this technology to transform the customer experience and maximize productivity. Generative AI, it can help you do more with less. And a lot of you agree with that. In fact, 84% of leaders like you, they believe that generative AI will help their organizations better serve customers. But Contact Centers, as you know, are struggling to transform with artificial intelligence. Today's contact centers, they're really actually looking at ways that they can go ahead and reduce those high cost and high volumes. And if I was to go ahead and Google your company's name and your customer support, I'd probably get a phone number. If I called it, the experience might actually not go as well as you had hoped, because your poor agents, they are swiveling between applications and trying to find the answers as fast as they can to deliver this great customer experience. And AI presents this unique opportunity to transform the legacy contact center into an intelligent engagement center. It will allow you to lead with intelligent experiences through self-service and digital channels, which generates tailor-made answers to your customers' questions. Again, an intelligent engagement center will help you personalize experience by grounding the answers and data across your customer journey. And an intelligent engagement center will help reduce that swivel-chairing for your agents by creating this unified customer profile. But [ for each ] Intelligent Engagement Center, which is easier said than done. 76% of the service leaders say that they struggle to deploy AI effectively. And the other 24% probably aren't being completely honest. The primary cause for pain is due to fragmentation, fragmented applications and fragmented data, which leads to service teams delivering this fragmented customer experience. Now I'm going to pass it over to Oana, who will share how fragmentation can impact your company and more importantly, your customer experience.

Awesome. Thank you, Amanda. And nowhere is fragmentation more prevalent than when we're talking about your customer data. Most businesses that I talk to every day in my role, they report having lots of customer data, but just not in a seamless, unified fashion. They've got transactional data in a data warehouse. They may have their loyalty or engagement data in a specific marketing tool. They may have their e-commerce data in a different system than their POS data. All of this leads to an environment in which it is really difficult to deploy AI and specifically Generative AI because, again, you're not going to be able to create those really personalized experiences. So what do we do about it? I'll present a framework to you today, and this is a framework that can be leveraged to really get started in a very easy manner with generative AI. First, we need to build your source of truth. We talked about data kind of being spread out across different systems. With Salesforce, you can leverage Data Cloud as that point of unifying all of your various data sources. Data Cloud offers a variety of out-of-the-box connectors for batch ingestion, for streaming data as well as if you've got a mobile app that needs to stream data in real time to Data Cloud or if you've got transactional data that gets reconciled at the end of the day, that needs to be a batch process. You can bring in data from everywhere. You can bring in data from your cloud storage, such as S3 buckets or maybe you've got data in another data lake like Snowflake, all of that can be unified in data cloud. Once you've ingested all of the data, you -- what you really need to do is take a step to transform the data. This is kind of where you take a moment to figure out, is my data in the right sheet? Do I need to do any type of cleanup? Do I need to do any kind of formatting to make sure that the data is in the right format for me to move on to the next step. The next step is really harmonization of that data, and this is where you kind of create a graph between your different entities to really understand the various relationships between the data entities that you've brought in. And that will allow you to now create a unified customer profile. And that's kind of what you absolutely need before you can move on to creating those personalized experiences, leveraging this amazing part of Generative AI. Now the next step is connecting your digital channels. And this isn't just connecting your digital channels, this could be your voice channel as well. And keep in mind, these 2 steps within the framework can work in parallel. They don't have to be sequential, even though they are displayed in a sequential manner on the slide. They can be done in parallel. Or they can be done the other way around too. You can connect our channels first and then build your source of truth. The idea is you want to make sure that all of your channels are connected because not only are your channels a medium for deploying these amazing generative experiences that can amplify your agents kind of -- or augment your agents power during the conversation with the customer, but it can also be a source of data. Think of all of the amazing conversational data that gets generated when you've got your customers engaging with your agents on a daily basis. And then number three, as we move on, this is where you create your trusted Einstein experiences that are grounded in your CRM data and that are encompassed within our trusted -- trust layer. So with that, I want to make sure that I highlight, this is all possible through our Einstein 1 platform. And what our Einstein 1 platform really brings together is Data Cloud, our CRM apps and Einstein. And we do that by making sure that Data Cloud is kind of that connective tissue between the CRM app layer as well as your Einstein intelligence layer. So that sales reps can create -- generated or can take advantage of generated e-mails by Einstein that are grounded in your CRM data, all fed in by Data Cloud and really personalize so that it is meaningful for that particular customer. But we're here to talk about service and how you can scale your contact center. So let's double-click on service really quick. Well, I mentioned in step #2 the importance of bringing in your channels. You talk to any service leader today and one of the biggest kind of pain points is, that they have channels across the board. They're kind of spread out across different point solutions. It could be that marketing made a decision to buy some marketing-related tools that brought in a variety of channels, but really wasn't designed for service use cases. It's more about engagement, scheduling posts and stuff like that. How do we bring in all of these channels that are relevant to your business in a service environment and make sure that those are natively built in your agent desktop? And that's where digital engagement comes in. Digital engagement allows you to bring in all of the different channels, whether that be WhatsApp or Facebook Messenger or maybe is a first-party chat that can be leveraged on your website or can be leveraged in your brand's app. All of those channels can now be connected and are part of our digital engagement offering. They are natively integrated into Service Cloud so you'll be able to leverage those within agent desktop. And if you have channels perhaps in a specific geography that are not natively integrated in Service Cloud, we do have what's called the bring-your-own messaging channel framework, it is an API framework that allows you to connect in any channel, any messaging channel across the world. So it could be -- maybe you do business in Japan and Line is really prevalent there. You can connect in Line, you can connect in KakaoTalk for free or maybe you have a big gamer community that you need to connect with. You can bring in Discord, for example, and connect that with the BYO APIs. The long story short is, no matter what channel you need to provide service through, with digital engagement, all of those will be present within the platform. And because digital engagement is connected into Data Cloud, data flows in and out of Data Cloud and can create those personalized experiences whether those are being delivered by a human agent or by a bot. So with that, I do want to show you kind of 3 ways that we can scale contact center, leveraging Generative AI. And we'll tackle 3 different personas here. So let's start with customers, right? Let's walk together through the customer service journey, a customer may be searching for answers on your website, right? They might go to your website. They have a question about a specific product or experience that you offer. And the first thing they're probably going to do start searching, clicking around trying to get some information. With generative answers, we are able to surface generated content for your customers that are grounded in your trusted knowledge base. So this means that the answer is that Einstein is generating for your customers as they're searching are not just super creative answers that LLMs can naturally provide, they're grounded in your knowledge database. So they are being presented with relevant factual information from your knowledge database. With Einstein Bots, another way of engaging, if they haven't found the answer that they needed on the website, they're probably going to be presented with a chat window and a bot may offer service if there's something else that they may want to engage in. So again, with Einstein, you can deploy these very rich interactive kind of chatbot experiences. We're talking about generative bots now. We're no longer talking about that deterministic flow that you used to set up back in the day with deterministic bots. These bots are much more powerful, and they can leverage the data that you've aggregated in Data Cloud to deliver super personalized experiences that are meaningful to your customers because the bot can now understand exactly all of the data elements about that particular customer. Now we'll take a look at what happens when maybe the bot couldn't answer all of the customers' questions or maybe it's a specific task that needs to be performed that your organization has decided a human agent has to take care of. In that case, this would get escalated to an agent, the agent would engage in the conversation. What happens now is that all of the data, remember when I said earlier that this is a 2-way flow, it's bi-directional between Data Cloud and Service Cloud, digital engagement is that data is now getting generated in these channels, as you're having conversations with customers and all of that's flowing to Data Cloud. What that means is that Einstein is now fetching the data from Data Cloud being able to analyze that and is generating these replies, service replies on the right-hand side there, and we'll see that in a bit in the demo as well. The agent now has the option of deploying these in the conversation. And of course, any engagement from the agent with these replies whether they are deploying them in the conversation, whether they're editing them before they're sending them, or if they're not using them at all, all of that is feedback that makes the model better and better going forward. And again, these replies are grounded in your knowledge. That's the most important thing that I'd love for you to take away. These are not just random answers that the LLMs are coming up with. They are grounded in your knowledge, meaning that the answers that are going to be surfaced are going to be factual from your knowledge base. And if your knowledge base is not sitting in Salesforce, can bring in an additional knowledge base, that means setting your SharePoint or maybe it's on our website, it could be in a KMS system. With unified knowledge, you can bring in that external knowledge database and make sure that Einstein is grounded in that. Another feature that I want to highlight here is, again, if the agent resolved the issue and maybe it's -- your agents are full of information. They're problem solvers. They're experts at what they're doing, a lot of times, they don't have enough time to create knowledge articles from what they're doing on a daily basis. So what this does with knowledge generation is anytime a case is being resolved or a conversation has happened, we're able to generate a knowledge article for that -- from that conversation, especially when something didn't exist when you have a knowledge gap in your database, this is tremendously useful because now you can institutionalize that knowledge that's in your agent's head so that another agent, either junior or somebody new to the organization can then leverage that knowledge going forward. Work summary is tremendously helpful as well. Think of the areas that your agents really don't enjoy doing on a daily basis, right, Case Wrap-Up. Sitting at the end of a case after you had a great conversation with the customer and having to now write and summarize the entire conversation, it's pretty painful. So we're going to do that for your agents. We're going to do that at a point in time. We're going to do it at the end of conversation. We have point-in-time summaries, meaning at any point in time during the conversation, you can generate a summary of everything that's happened to date, so that if you're transferring that to another agent, that agent can now see all of that context that very quick summary as they're accepting the transfer. And then at the end of the conversation, obviously, we're also going to be able to summarize that you can store that on any object that you want. If you want to put that on a case, if you want to put that on a custom object, the work summary can be placed anywhere. All right. So let's get to data and insights, right? Another way that we can make sure that you can scale your contact center, is by looking at your contact operations. We want to know how your agents are doing. So AI can now be leveraged to generate surveys. You can generate a survey to make sure that you're gathering that feedback about your agents' performance. And then kind of the other end of that is service intelligence. We can make sure that you are really connected to all of the data about your contact center. We have these amazing service intelligence dashboards that you can leverage to look at how your contact center is performing. And that's super important for managers as they seek to optimize and make tweaks in their operations and processes to optimize their contact center. So with that, I want to leave you with kind of one basic idea is, it's really not intimidated to embark on the Service AI journey. Think of it as 3 steps, right? Think of how can you decompose the service journey for our customer and really dive deep into those things that can be automated? What are those processes that are repetitive that can be automated either by a flow or by a bot? And then if it can't be automated and it has to be done by a human agent, how can you assist that agent? So those very kind of manual tasks that an agent has to do can now be assisted by generative AI. And we can relieve some of that pressure off of your agents, so they can do what they do best, which is really engage with customers. And then optimizing your contact center through some of these amazing tools like service intelligence, which can be used by your supervisors and contact center leaders to optimize your contact center. With that, I'd like to present Tamer. Tamer is our Principal Solution Engineer, and he's going to walk you through an awesome demo, what I just talked about.

Perfect. Thank you so much. Give me one moment. I'm just going to share my screen here. So everyone can see all the great stuff that you talked about. All right. So my name is Tamer Farag, a Solutions Engineer Salesforce, and I'm looking forward to walking through the presentation today about really how everything we want to talk about and how we can really bring that to life. So in today's demonstration, we really have 2 characters. We have Rachel, who is a banking customer, who's having an issue with her debit card and essentially needs it replaced. Then we're going to have Tim who is an agent in the contact center, and we're really going to show you how all these different type of tools can help Tim on his day-to-day job. So having said that, let's actually get right started here. All right. I just want to confirm, you can see my screen now. And what we're looking at right now is essentially the search answers. So going on the actual website itself, in this example, it's an Experience Cloud, I have to be able to search for different type of issues. And what this is going to do is actually bring up generative type of answers here. And it's also going to be grounded with the knowledge that is behind the scenes on how to do that. You can see here that Rachel has gone ahead and searched indicating that she's having issues with her debit card. In this example, her tap is no longer working. So a lot of different things come up here to be able to help Rachel be able to self-serve that. But at the end of the day, she actually got -- needs a new card. So she can easily do that and go through a bot to be able to handle this type of situation. But in today's demonstration, what I want to do is, I want to get into the digital engagement and the actual agent experience. So having said that, she selects web chat and actually goes through directly to an agent itself. Right now, what we're looking at is at the agent desktop. So we switched over to Tim. And this is essentially ways that your agent can actually go through and see different type of information here on a day-to-day basis. You can see different things here is that they have on their home screen. How is the contact center doing? How am I doing with different type of scorecard? You can see that my status here right now for omnichannel is, I am not logged on. So essentially, all the digital type of channels, I'm just getting started from a day, and I'm going to go ahead and log on to that. As I do that, you can see here that all the different type of channels that I want to talk to, whether it's messaging, whether it's chat, whether it's the phone channel, WhatsApp, all these different channels, we digitize them and they all come into Salesforce to be able to help your agents. So they're not swivel-chairing to different applications to be able to support your customers on their channel of choice. In today's demonstration, I actually want to show you chat and show you how that actually comes through. So you can see here now that the chat is coming through. As an agent, I've gone ahead and accepted that chat itself. So a lot of things are happening here. Very first thing is because Rachel was authenticated, and I know who she was, I'm actually able to see her information and utilizing things such as service intelligence or service insight, I can see the type of customer Rachel is. I get access to different things such as her CSAT over time. What has been her sentiment score? What about that? Is she had her NPS score, Is she a promoter of us? Does she currently have any type of issues that I can help her with while I have her on the phone? Other things that we can really take advantage of since we have that, is what about Data Cloud? All that information that's harnessing for all your different type of data sources we can pull that in, whether it's stuff that you're doing on your website, perhaps browsing, you're looking at different things, we can pull that in. All of this information can provide you and your agent a lot of different things for that. So just providing that and pulling all that information unifying and harnesses, that will help your agents be able to help Rachel for that. Now we're actually looking at the conversation in the middle of itself. And this is really what's really kind of the need of everything we're looking for. And I'm going to go ahead, I'm going to respond to Rachel, how can I help you today? And she's indicating that her card is not working. Over here on the right-hand side, you can actually start to see service replied. This is actually going in and getting generative answers on how we can actually help out Rachel for that. You can see here that these answers are brought in, and they are also grounded with our trust layer. So they have different type of abilities, looking at knowledge, looking at past cases, to be able to ground that information and be able to trust that information for her. So I'm going to go ahead as an agent itself and just go ahead and post it. You can see here that I also have the ability of editing it or indicating that it's not helpful, which again, trains that actual engine to be able to help our agent. In this particular case, you can actually see that Rachel's indicated that her tap is actually not working on her debit card. But looking at that and because we actually have the context, we're able to drive a different type of workflow behind the scenes. So different things such as next best action, which can drive different type of workflow. In today's example is to send out a new debit card. So instead of sending an e-mail or flipping to another type of application, I have the ability of triggering different type of workflow that could go to the back office and send that out. So very quickly, I can click on that, and we can send Rachel a new card on that. What I'm going to do is I'm going to ask them if that's okay, if we go ahead and do that. And you can see here, looking at the various type of response, we're again troubleshooting with there, and we're going to go ahead and send her the new card itself. She's indicating that she wants to be able to do that. So we're going to say perfect. We're going to go ahead and send one. And then we're slowly going to wrap up this particular instance. But again, you can see all the service replied to be able to help Rachel for that. Now that Rachel is going to receive a new card and everything is perfect. How can we actually help the agent? Agents spend a lot of times indicating different type of information, a lot of times and wrap up, entering notes after a particular conversation. By leveraging different things such as work summary and case classifications, we help that agent do that. We're simply looking at the transcription here of what just happened within that chat. And by putting in the recommendation itself, you can see here, it's indicating everything that's happened, it's summarizing that, indicating the issue any actual resolution all automatically from looking at that conversation using Generative AI. As an agent, I can review this. Everything looks pretty good. I go ahead and actually save the details itself, which is actually now saved in the actual record itself. So when you kind of recap really what just happened, Rachel was on our website, she utilized the self-support. She really went through the search answers, but then realized that, you know what, she needs a new card. Her card is not working great. From there, she utilized her channel of choice in this example, it was chat. That got pushed through to an agent. The agent accepted that particular chat, had that complete customer view of Rachel, had all the different tools available, leveraging tools such as Data Cloud, generative responses, generative work summaries, next best action to be able to help Rachel. So you can see very quickly by leveraging this technology and everything that I want to talk to, we're able to make that agent's life so much easier by actually be able to do all these different things for that. But what about making sense of all this information? How can we do that? By using service intelligence, you can start to mine all the different type of information. You could look at different things such as conversation line. What are people talking about? What's going on within the contact centers? Utilizing this information, you have the ability of adjusting your contact center, making any type of adjustments such as, do you need to train agents on a particular topic that seems to be coming up a lot. In today's world, interest rates are always rising. We all know that, and that's probably a hot topic when people call into contact centers and in this particular example, a bank. So perhaps we should have a knowledge of, how do you handle customers' questions when you have questions on that. All of this is able to help your agents by utilizing service intelligence. And again, all this information is available from Salesforce, and you can see all of this available on our website. From here, what I'm going to do is I'm going to pass it over to Robin from Metrigy, who's going to actually go over and talk about some of the statistics. And with that, I'll give it to you, Robin. Thank you.

Great. Thanks, Tamer. And what a great demo. I had the fortune of being able to see this live at Dreamforce, and it was just feels so good. So hopefully, everyone enjoyed that. Yes, I'm Robin Gareiss, I'm CEO and Principal Analyst at Metrigy. I'm happy to be here with you today. I always encourage people to connect with me on LinkedIn. There's my name. I always post a lot of research data and things like that, that might interest you. So let's get started. I think that both Oana and Tamer talked a lot about what we see happening today and where some of the need might be when it comes to using advanced technologies like AI and Generative AI with your customer experience strategy. Well, why do we need to do that? What's really driving that? I would say one of the big, if not the biggest thing, is you want to give your customer a good experience because if you don't, guess what happens, you're going to lose them. Three strikes here out is kind of the mantra here. We -- in a consumer-based study that we just did a few months ago, more than 500 companies -- consumers rather, we ask them how many bad experiences, first of all, would it take for you to leave the company? And on average, it was 3.3. But you can see from the green bar chart on this slide, what do you do when you have a bad experience with the company? You just want to know what happens. 44% of them right after that say, "I'm done. I don't give them any more business." So to at least 44% of consumers rather, you just have one chance with them. So when you think about how consumers are reacting and what their demands are, it's kind of crazy. They're not going to give you much leeway. Other groups are going to want to have a megaphone. They might stay with you, but they're going to tell their friends and family about their experience. They're going to leave negative reviews on websites or post on social media. And that also has a really bad effect because it may prevent you from getting new customers. It may cause existing customers to say, "Oh, maybe it's time to leave." Another group that might do what you'd like them to do if they have a bad experience. And that is talk to a supervisor, send an e-mail to customer support, even just try to work with the company to improve. And I think that with the technology that we have today, we can push more people who have bad experiences to those areas where they're going to help and they're going to work through because we can use things like AI, like sentiment analysis, like inferred sentiment, all that, to determine when somebody might be at risk of leaving and be more proactive and reach out to them and try and get them to work with you to fix the problem, tell them that this is a concern and that you are going to fix it and they're going to help -- if they can help you, that might really -- it might give them the right impression of your company. So I just want to make that point that you don't have a lot of chances to do the right thing for your customers. So let's figure out what we need to do. All right. So one of the big things we obviously see companies doing already is adding all different kinds of AI. I mean there's so many different types of AI. You can be doing transcription, translation, virtual assistance, virtual agents, sentiment analysis, predictive, all sorts of -- there's no shortage of types of AI that you can use in the contact center or with your CX strategy in general. What we have found is that using AI is affecting your hiring. And let me explain. I get asked all the time, is AI going to take my job? I will tell you, and you'll see more of this in the next slide. We are not yet seeing that AI is taking existing people's jobs. We are, however, seeing that it is reducing the amount of new agents that companies need. So we asked about this in our research in June. This is almost -- I think it was 641 companies. And we asked them about their use of AI and hiring. So how many new hires would they need? Were they planning to hire and we correlated whether they were using AI or whether they weren't, and for those who were using AI, we said, well, if you weren't using an AI, what would you anticipate you would need to hire? So we kind of crunched all those numbers. And what we found, honestly it was a little shocking to me. I didn't think it was this significant. But if you are not using AI right now in the contact center or with your CX strategy, you will hire 2.3x the number of new agents in 2023 compared to those who are not using AI. So you can kind of see the numbers there, the averages, the right-hand bar chart there shows you, it varies based on the size of companies, the raw number. But the percentage is about the same across the board. And when you correlate that with the annual savings, we asked everybody what they're paying their agents. We added a 30% loading factor and everything. Average annual savings is about $4.3 million just in that staffing compensation that you are now going to avoid because you're using AI. And when you add to that, or sort of consider along with that, how much companies are spending on AI right now in the contact center, again, those are just averages. They're going to be higher and lower based on the size of the company, but it's about $500,000 is the average. So if you compare spending $500,000 for AI and saving $4.3 million, again, just 2 average numbers, it's pretty compelling. I'd spend that all day to get that kind of savings, right? So the issue, though, is that we're not seeing this replacement yet of agents. So first of all, let's look at the data, about 40 -- almost 47% of companies say they're going to still increase their number of contact center licenses by 2024. Only 6.6% are decreasing very small percentage, 42% staying flat. And when you ask those who are decreasing, again, just a small percentage, why the top 2 reasons are very, very much because of AI. Customers get more answers with self-service, which is often fueled by AI, and AI is replacing agents. Okay, very small percentage are saying that. What I see happening moving forward. Will AI replace some agents? It could. But I also see a lot of job-shifting going on. I think we're going to need more people in knowledge management. We're going to need more people to look at the content that we have to make sure Generative AI is pulling from a very up-to-date, accurate multimedia database. And right now, I don't think we have enough people there. We're already seeing companies pair their agents, their live agents with a team of virtual agents and the live agents are kind of keeping those virtual agents on track when they go off-track. So that's another area that we're seeing. And in other cases, I will say that I talked to many companies where they were really understaffed. And by the way, this is the first year in all the years I've been tracking this, that the contact center is -- the majority of contact centers, 58% are fully staffed right now. A year ago, that was 48%. And that is not because people miraculously found all these new agents to hire. It's because they started using AI and it's really helped them to backfill. So I find that when companies are in this growth trajectory, they're growing and they need to -- they need more and more agents consistently because they keep growing and growing. What's happening is now they can handle a lot more with that existing field of agents, and AI is helping them handle that growth. So they still need some growth in agents but not as much as they would have without AI. So that's another very common scenario we see, and that will not cause layoffs. But in fact, AI is going to help augment those agents. Now one of the big areas of AI that we certainly see and that we've been talking about already today is bots, whether it's chatbots or voice bots, more often chatbots and voice bots, but we see both. And we asked organizations what percentage of your transactions do voice bots or chatbots, touch? It's about 38.5%. And of those, 42% were resolved with bot. And when you look at those numbers and think, Oh that's not too bad, but then look over to the right. All those little people there represent a case, a transaction interaction. Any of them in color were touched by a bot. Those in orange were resolved with the bot. Now it doesn't look so great. Only 16% of interactions are actually resolved with a bot at this point, even in our research success group, which are those who are doing everything right when it comes to business metrics. They're above average in all their business metric improvements. They're only at 19%. So we still have a ways to go when it comes to bots and the success rate of the bots. I think we're absolutely getting there. I think generative AI is going to help a lot. I think vendors are coming out with a lot more purpose-built chatbots. I think early on a lot of companies, they didn't train their models well. They over and misapplied their bots or they have them doing functions, they really shouldn't have been doing. And that caused a lot of really bad start -- kind of a bad start to bots, but I feel like we're kind of turning around now. In fact, when we asked our consumer audience of what their most recent chatbot experience was like, 44% actually said positive and only 20 -- I think it was about 23-ish percent said negative. The rest were kind of neutral. So that to me shows we are starting to kind of turn around because when you look at the number of companies -- our consumers rather right now who want to use a chatbot, who prefer a chatbot, it's only 13%. Now 47% will use them in select cases, like they know when a chatbot is going to do for them what they needed to do, and they'll be happy to use it at that point. But it kind of depends what the use case is. We are, however, seeing that bots are helping in AI and specifically virtual assistants here, in certain agent use cases. So right now and toward the end of COVID, we really saw this number going up. The average number of simultaneous chats per agent, prior to the pandemic, it was 2. As the pandemic went along, it got up to 5.8 and that's where it stays. It's hovered this, at this way into for 2 years now. And imagine being an agent, having 5.8 simultaneous chats, right? That's a lot. And obviously, your customer isn't going to get that real time kind of back and forth. There's going to be some delays, we have all probably experienced it. Well, what companies are doing is about 38% already are pairing the agent with a virtual assistant, 37% more were planning to do this by the end of this year. And we see a lot of really good results with that. The agent assist will help the agents themselves resolve questions and inquiries faster by giving them context and next best action recommendations and so on. They also will help customers by opening and closing the chat maybe in between doing some automated functions. But when we found that when companies paired the virtual assistant with a live agent, the agents were able to handle almost 8% more simultaneous chats. Not an earth shattering number, but certainly a measurable number. And I think that because companies are doing that, we're not seeing that average simultaneous chat number drop even since the pandemic ended, where we see a lot of our line chart numbers kind of changing now. Because of AI, we're seeing agents being able to do a lot more, very interesting use case there. Now the other area where we really see agent assist helping -- I know we're talking more about service today, and this is related to customer service representatives, but now 54% of companies actually have sales quotas attached to their customer service representatives. So these aren't your inside salespeople who also, of course, have sales quota. These are your service reps who may not have that sales experience, but now they have quotas. They might need to ask for upsell, cross-sell. You want a warranty? What about this product that goes along with that one? Some are comfortable with it, some are not. But when you use agent assist, that can really help getting those [ screen helps ] to kind of give you a little push and let you know what might be likely for this -- what's likely that this consumer might buy, that's really going to help you do a better job and meet some of your sales KPIs even as a service agent. So when customer service agents who have these sales quotas, also used agent assist, their average handle time actually dropped by 28%, which was about 9% better than those not using agent assist. So we really definitely see any sort of AI coming into play, just helping improve business metrics across the board everywhere we look. And as a result, by the way, average handle time is, in fact, becoming a less important KPI than it once was. We used to want to get people on or off calls, as quickly as possible to get to the next one, keep our cost low. Now, it's like, well, if you stay on for an extra 2 minutes and you sell a new warranty plan or you upsell somebody to a new product, that extra 2 minutes is well worth of time. So that's some of the things that we see companies doing with AI on the sales side with their service agents. Another area I wanted to touch on is, and Oana I think touched on, how we can now do AI-enabled surveys. And we asked companies how they gather customer feedback. So this is where we're getting that input from customers so, so important. Without it, how do we know what we're doing is working or not working. I mean it's vital to have that feedback. So you can see the #1 way that companies are getting feedback at this point is through non AI-enabled surveys, just through typical post-call survey, proactive survey. But we also see all different types of feedback methods, one-to-one conversations, getting feedback on social media. Just looking at your website statistics. But you can see there fifth bar down is AI-enabled predictive surveys. So these are, of course, letting Generative AI create a survey, but also letting it look at and read the responses, especially open-end responses, coming back and noticing trends even before people do and then they can actually alter -- AI can actually kind of alter the questions or add some branching questions to a survey to get more information, to get more of the whys behind the what. So really super cool technology that you should be looking at because it really helps you to gather better and more detailed feedback from your customers. But other ways companies are gathering data, third-party aggregation sites like you've got your Google Rankings on hotels.com and things like that. Live chat, so just doing a chat with somebody. Focus groups are another big one. AI analytics, I think this is going to really go up in the coming year. This is where we're looking at sentiment analysis and NLP and even doing some inferred sentiment, which is, if my customers aren't going to send me any response or only -- I mean, honestly, on average, it's about 30% of customers actually respond to surveys. And again, this is an average, it might be higher or lower. But wouldn't it be great to have a response from 100% of your customers? You can with infer sentiments. This is where AI is listening to your conversation, detecting your sentiment, listening to your word choice, seeing what the end result was and comparing that to a large AI machine learning base that has other conversations and comparing yours to ones that might be similar to yours, and where they did get live responses and say, well, looks like this call was very similar to these over here, and we're going to infer that if Robin were to answer this survey, she would have given our interaction an 8. And by the way, those inferred sentiment capabilities are incredibly accurate. Early models of these are running in the 85% to 90% accurate range. So it's much better to be able to take that feedback and not just based what you're going to do with your agents and coaching and technology and all that stuff, on 30% of your customers, but based it on 100%, AI is able to do that. It's really cool. IoT and embedded devices and kiosks are other areas where customers are -- or where companies are gathering customer feedback. One thing I will say is our research success group is using more of these channels than the nonsuccess group. I know the numbers on the right don't look hugely different, but when you consider there's not that many methods to use, it statistically is fairly significant. So you want to make sure you're using at least 4 different types of methods to gather customer feedback to gather very well-rounded feedback. And try to use different methods. So maybe you're going to use a survey maybe a focus group and then some social media and website stats. You can see how you gather different types of data if you did that. Okay. The other thing that we see when we gather that insight from customers and we have AI analyze the results, it really is helping to fuel sales growth. So again, I know we're talking more about service here, but let's look at what we can do with insights and sales. We asked organizations, once they got this insight from customers, you have AI look at it, especially for your qualitative response. Yes, you can look at your 5-star or 1 to 10 ratings very easily, very quickly. That's a real quick analysis. But when you've got open-ended responses, that has some of the most valuable data contained within it. And you can't possibly read it all as fast as it's coming in, AI can. And it can take that information and get it right over to the people who need it to do immediate improvement in whatever their problem areas may be. So you can look at these really compelling numbers. When sales teams started customer insights programming used AI to analyze the results, they were seeing increases in productivity, deal sizes, pipeline movement, revenue, close rates, [ ramp rates ], everything here got better. And the customer churn rate, by the way, was a decrease. We just put it on the chart, so it didn't make the chart look funny. But it's seeing that much kind of improvement in those numbers that really matter to sales. This is where AI can come in and help again. The problem that we see in general right now is that AI may not be fast enough, it may not be able to do as much as we need it to do, in terms of reading that content like those customer insights we just talked about, this is where generative AI is changing everything. And you can see here, as of June, this is when we gathered this data, 27% of companies already were using generative AI. I wouldn't say they were doing anything incredibly sophisticated, at that point, it's more like kicking the tires, doing some summarization, some content creation, some classification, things like that. Another 47% planning to add it by the end of this year. We are going to -- about to launch another study here in the next 2 weeks, where we're going to update a lot of our AI data, including Generative AI. And normally, we wouldn't do that -- we kind of look at this annually. But in this case, we're doing it at the 6-month mark, just because it is moving so quickly, we want to understand the ROI use cases. So that will be coming out in just about 2 months, we'll have the data ready for that. But you can see that there's a lot of adoption -- very quick adoption, of course, for sure. And the problem with Generative AI, I mean you may have seen some news yesterday that the White House issued an executive order to do some -- to put some regulation on AI, to make sure that companies are testing it. I'm not a huge government regulation fan, but you certainly can see some of the concerns that people might have about Generative AI. Especially when ChatGPT first came out, there was a lot of doom and gloom stories out there, it's going to end the world. It's going to take your job, it's going to start a war, all the stuff that Generative AI, they were concerned it would do. We didn't hear a lot of positive stories about it. So I think that, that really has shaped a lot of the public opinion right now in generative AI. You can see we asked both the IT CX and business unit leaders in one side of the study and then consumers and the other, what would make you trust Generative AI more? You can see on the left-hand side, limiting the data it can use to create content. So having guardrails around that data. Businesses understand that they're like, yes, this is what will make me trust it more, consumers, it doesn't do much for them. What they'd rather see is human oversight to its content creation, which we also see the business side looking at and limitations and its capabilities. Again, those things may in fact be necessary, especially early on, but it does stifle, slow innovation and creativity. So you kind of have to balance those 2, for sure. You can see on government regulations, not many either business unit leaders or consumers really were keen on government regulations. As I interviewed companies for this research, a lot of them actually talked about wanting more of a standard body type of environment where business -- where the vendors, some like technology leaders from companies like CIO, CCOs, and even government and academics all came together and sort of did things more on a standard body type basis as opposed to government regulation just because oftentimes government regulations slow things, and that's probably the concern there. But what's really important here is that consumers are saying they will never trust Generative AI 4x more often than the IT CX and business unit leaders. It's kind of crazy. And what I think we need to do here is a lot more education, exactly what we're doing here today. Let's get out there, let's talk about the technology but show how it can be used. Let's get people comfortable with it. And it's not to say that there aren't concerns. Believe me, there are concerns. There are definitely bad actors out there. There is definitely concerned about malicious intent. But we have to really balance that on what it can do on the positive side in terms of productivity, in terms of health care, like that's a big one, how can AI help health care. I mean it's interesting. My daughter is a veterinarian, and she was telling us just recently the AI now reads their X-rays. Now you still need that human oversight to that because sometimes AI misses things. But she said sometimes AI also catches things that the humans don't catch. So it's kind of a partnership. Again, I don't see -- and she doesn't see AI ever replacing humans looking at those x-rays, but it certainly is a partnership. And I think that's what we're going to see in all walks of life, in all areas business. It's going to be a partnership, and we're going to have to really work to make sure that we've got a good balance going on. So wrapping up, AI is a real-world impact. A few things I want to leave you with AI absolutely now yields significant savings by reducing hiring plans. It's not necessarily cutting jobs. We do see a little bit. And I guess if you look at future jobs, it might be cutting some future jobs. But honestly, everyone has such an overhang of open [ recs ] for contact center agents that they could not fill. It's not even like you have, you're really cutting jobs because they were jobs no one they weren't filling. They just couldn't find enough people. So now we're getting this good mix of AI and human really managing things well. And it also, of course, boosts the efficiency by helping agents, what is it helping them do? Handle more chats like I talked about, but also giving them better analysis of customer insights, which ultimately will improve sales, improve revenue. using AI to read those open-ended comments and taking those and do something with them. And then finally, Generative AI, the adoption is really strong right now. But hesitation about the sensitive data and how you can make sure people trust it, that remains the top barrier right now for trust. I think we need to address that and educate people about the value of the technology as well. All right. So with that, I think I'm turning it right back over to Amanda for wrap up and perhaps some Q&A.

Yes. Thank you so much, Robin. Three strikes in your out, has a new meeting for me. So, thank you so much for the very informative walk-through. I want to just wrap up and maybe we can take a few questions. I know we're almost at the top of the hour, so I'll be very speedy. But I wanted to leave the audience with 3 resources to get started with Generative AI. So first, if you could check out the blog by Ryan Nichols, he's the Head of our Product Management here at Service Cloud. Ryan highlights the 4 ways to get started with Generative AI in the contact center. Next, you can learn how to lower your cost in the contact center and deliver that fast, empathetic service at scale by putting the customer at the center of everything you do. It's a nice HBR article. And then the last thing, if you're interested in scaling faster, checkout Service Cloud Unlimited Plus. Unlimited Plus offers the best of Salesforce for service combining trusted AI and harmonized data with a market-leading CRM for service. Only Unlimited Plus brings together Slack collaboration and robust analytics in Tableau into one solution for you, helping you build for today, and also plan for tomorrow. So if you check out that demo link here on the slide, you'll be able to see a demo of that and also be sure to connect with your account executive.

So I wanted to quickly just ask some questions we've received. First one, Oana, is for you. So you were talking about service replies for agents, and you showcase the chat example, customer Michael had a question around, does this apply to all channels, any channel like phone calls?