Scaling the Agentic Enterprise | Salesforce on Salesforce Keynote, Dreamforce 2026

Please welcome President, Enterprise and AI Technology, Salesforce, Joe Inzerillo.

All right.

Good afternoon, Dreamforce. I'm just going to say right now, I think we may need a bigger room. But thanks for coming out. Thanks for seeing this. It was an amazing keynote from Mark. But right now, what I want to do is I want to take you inside of Salesforce. I want to show you how we use technology, some of which we build, some of which we don't build, but how we use it to actually run the company. And so hopefully, some of you attended the keynote last year. And this was sort of what our theme was, which is, how do you get started with agentics? And so start small, go fast, test, tune, and repeat. The right data is gold. Measure the value and the impact, and embrace new ways of working. And all that is very true, especially if I'm started your journey. But today what I want to talk to you about is how do you scale that? How do you take it to the next level? How do you deliver incredible impact by doing things to make yourself an agentic enterprise start to finish? And so this is what scale looks like. So when people say, what do I mean by scale? What I mean is 100% of our employees using agentic tools. It's a huge, huge thing. 200,000 Slackbot skills created just in Salesforce. That's not our customer skills, that's not your skills, that's just inside the company. And there's $160 million of pipeline from a sales standpoint that wouldn't exist had we not become agentic. And I think it's an incredibly important thing, but I wanna spend a second and I wanna talk about agent scale. And so when I say agent scale, what does that mean? Well, on the other side of the scale, we have human scale. What can a human do? How do humans work? They go to a piece of software, a web browser, click around, do menus, things like that. Maybe they go to Slack, they type to other humans, and they do this. And so you can be better at it or worse at it, but there's a finite amount of scale you as a human can deliver using the tools that you have. Now, agentic scale on the other side of it is things that a human could not do, things that are incredible feats, you know, 3 years ago were science fiction that we can do today, like reason across every health conversation that's ever happened, or search the corpus of 27 years of Salesforce data and give you an answer. These are agentic scale. And the thing about agentic scale I'd love you to hold in your head is that if you take something and you make it agentic scale, the thing that comes after it also has to be agentic scale. If you don't, and that's human scale, that is where the ROI goes to die. So when you hear, oh, 95% of companies aren't getting the value, they're not getting the value because they are not making the entire process agentic. They're only making a portion of it. And that localized optimum is just not going to get it done. So how do we do it? How do we get there? Well, Mark's keynote did a really good job, I think, of explaining this. So I'm not going to drain the slide, but The data layer, the customer layer, the applications that you know and love, now often in a headless variety, sitting at that Customer 360 layer. Agentforce, the agentic layer mixed in there too. And then AI Force, which I think is just one of the most exciting things, is this new interface, the new way of thinking about interfaces, thinking about workflows. And you're going to see a lot of that in this demo. So The thing I would like to start with though is just giving us a framework to walk through these. And so these 5 things, if you're gonna take anything from this keynote, these 5 things, or at least some of the 5 things, you don't have to get all of them, but hopefully you get 5 of them, is everybody can build. Not everybody does build. We used to say everybody's a builder, but what I've come to realize is that's scary to a lot of people. You know, when you talk to your business users and you say, well, yeah, but I'm in finance and I use Excel and I do this and that, does that mean I have to be a programmer? It's like, no, you don't have to, but everybody can build. Everybody is empowered to build, I think is critical. Again, getting back to this agent scale, how do you take a process and then redesign it so that it can work at agent scale? What are the right platforms? What are the right tools? How do these things fit together so that you can empower the humans? Then you can't just stop there. You have to think about, well, how do I rebuild my business? How do I rebuild the way that my org chart looks? How do I take advantage of this? Because it would be weird to try to do a technological revolution that completely changes the way we work and then say, yeah, but don't change any of the org chart. Do it everything else process-wise the same. So you have to build for the agentic era. Then how do you make— when you talk to your customers, how do you make that all one seamless experience? So the downside of the agentic revolution is that once you do something that's awesome, people expect it to be awesome everywhere. And by the way, when your competitors do something awesome, they expect you to be awesome too. So if you have a great experience with an agent, you're going to bring that with you to your next agentic experience. And it's sort of like For those of us who have cars that maybe have voice control, like back in the day, oh wow, look, turn on the stereo, do this. Now if you ask it the weather and it goes, I'm sorry, I don't understand that, you're like, this thing sucks, right? So you have to scale it to the whole journey. And then you have to build a future-ready workforce because the machines don't yet build themselves. So people have to build those. And so how do we think about reskilling? How do we think about our workforce? And we're going to dive into that too. But I want to start with That everyone can build, and again, the can is super important. In reality, most do. So when you look at folks inside of Salesforce, 67% of all employees are using a coding tool. Does that mean they actually know how to write code? In most cases, no. But they know how to talk to a machine. In our case, most often. through Slack and say, this is what I need. Here is a description. Here's my opinion of what I'm looking for and how I want you to go answer this question. The result is coding in many cases, building an ephemeral dashboard, answering a question, but they can build. And I think that's really important. The 57% of non-technical employees that are using coding tools is up 20% from Q1. So this is accelerating massively. And I think it's a critical, critical thing to remember that it will continue to accelerate. There is no end on this. It's going to keep going. And so we really wanna empower that. But to do that, you say, okay, great, everybody can build. First off, that sounds like a nightmare for a CIO. Everybody's building all this crazy stuff. And to some degree it was when we first started this process. But I think that's because when we think about the model in our heads of how IT builds, We think about it as we used to build the whole thing. So we would have, you know, product people, program managers, etc., that would go talk to the business people and say, hey, we want to get some information so we can deliver something for you. And that took time. Used to work waterfall, then agile was better, but you're still interviewing, building, and there's a delay there. But reality is that if you want to take advantage and become agentic, you have to think about this in the context of what is IT going to look like in the future? And what I feel like is IT is still going to very much be concerned with things like trust and integrity compliance. Like you can't vibe code your way to SOX compliance. That's not quite cool yet, but you got to make sure it happens. And the IT will have a very heavy hand in systems of record. Like what does our data look like? What is the shape of what a customer is, etc.? We'll also look at data semantics. So when we define things like what you know, AOV is or revenue. Like, that can mean a lot of things to a lot of people, and you really want folks that are hyper-specialized to figure that out. But I think that all gets started— starts to get conformed to APIs and MCPs because we're no longer going to be building for people. We are going to be building for agents that people are controlling. And so this API then allows the bulk of the work where all that business value is, that trapped value that Mark talked about in the keynote. That value is by getting the business folks to be able to do the coding themselves in a team. And we're gonna go into that in detail shortly. But we didn't get here overnight. It's not like we're like, oh wow, we're so smart. Look at us. We made 100 mistakes so that you didn't have to when we went on this journey. Some of those got expressed as products, which I'll talk about in a second, but a lot of them were just like, we know token maxing is probably dumb, but we also know we have to enable people, so let's actually go and try to do some of that. You know, we know that everybody's going to build a different way, and that's okay too, because we want to allow for improvisation. We also know that access is going to be an issue. This one we didn't really take lightly. This is one of those things where we started more modestly and then expanded as we got comfortable with Understanding that trust and security are critical for IT. But we also were thinking about human-scale limits. How do I affect the system as a company so I can take that human-scale stuff out and get to agentic scale? And the thing that we noticed most was teams wanted to collaborate. When you started to pull down these barriers, they wanted to work together. An example of that is This is expressed in Slack. So we just announced a couple of weeks ago, Mark talked about in the keynote, Slack Code, which is where people, technical and non-technical, can collaborate in a coding channel and talk about what it is we're gonna go do. Critically, critically important. So I wanna dive deep on that because that— I talked about a lot of technical change, but I want to invite up my friend Andy to talk about the people change part of it, about how IT remodels itself to make this happen. And when I do, let's go take a look at a little sizzle reel of what it looks like in action.

All right. Hi, everyone. I am grateful to be here today. My name is Andy White. Thank you, Joe. And I'm excited to share how Salesforce is transforming from the inside out with all of you, our customers, partners, and trailblazers. For a long time, every team, every process, and every system has been built around human scale, and that was the right thing to do until it wasn't. And now our opportunities and our constraints have changed, and it's time for us to change, to rethink the new world. What I'm about to describe is going to sound like Lombard Street here in San Francisco. You start with requirements, And then you wait, and then you curve around, and you have a feedback meeting, and then you schedule a design review session, and then you wait, and you have an architectural review, and then you go back to the original people that gave you the requirements, and they say, I want it different, and that's a change request. And so we said, what would it look like if we could rebuild our teams, our processes, and our systems for agentic scale? And how much faster could we go with humans and agents working together? So in order to answer that question, we created a time machine to the future, and it's called Rapid Innovation Squads. We wanted to understand what work would be like in 12 to 18 months, and then bring that back now and help us to change more quickly. And so Rapid Innovation Squads is a bit of a mouthful. We call it RIS. It sounds a lot cooler. And so we started these RIS teams, and they started with a really simple process, an opinionated stakeholder. And you have to start with the problem. You have to start with what are you trying to solve? And then we bring to the problem our agentic builders. And this could be 1 agentic builder or 2 agentic builders or 5, but it's as small as possible. And you wanna shrink the problem from the opinionated stakeholder to as small as possible. And so that all makes sense and is straightforward. But the real difference is in how the team works. The RIS team gets together in the morning, they drink coffee, they join a standup. That meeting is recorded and transcribed, and it's fed into a public Slack channel. We want it to be public so that the rest of the organization can learn from what's happening with this one RIS team. If you're working on something double secret, it could be a private Slack channel, but we think this is really important. And then you feed that transcript into an agentic coding tool, and you walk away, and the agents go to work on what you discussed. And then you repeat that after lunch, and everyone is a builder that's on this team, and they all show up with ideas and with prototypes, and they work together toward this outcome. And then they do it again one more time at the end of the day, and the agentic coding tools go off to work while they sleep. But the power of this different way of working is that you get 10x more faster decision-making when you're partnered with an opinionated stakeholder. And if you pair that with agentic coding tools, you can get 30x faster outcomes. But it's not just about speed, okay? It's really about the business impact that you're able to do in this new way of working. It's really incredible. And because these are a time machine, we're going to keep doing this and keep learning and keep learning. And a year from now, we'll have a different perspective than we do today. I talked about opinionated stakeholders, and I want to tell you about one. Our very first person was Maya, and she's in this room. She's amazing. I hope all of you have a Maya at your company. You'd be lucky if you do. And what makes Maia so special is she is passionate about the partner ecosystem at Salesforce. She is responsible for the partner experience. She is willing to disrupt that experience to move to something better and something new. And she's embedded in the rapid innovation squad that she's a part of. She's not just showing up with a prototype, she's pushing code to production. She's one of the 2 opinionated leaders that we're working with that has done that. And that's pretty amazing. And that's profound because she's not in IT, okay? And this is why Joe was talking about the importance of the IT team having the foundation with those Lego blocks that you can enable Maya to safely be a builder. She's really incredible. Now, I want to talk about a legal risk. This is case triaging, and I've got a bunch of stats that are up here behind me about faster deal cycle, about cost savings. We also are improving the seller and legal and customer experience through this, but it's pretty incredible, and it applies to every company that has a legal team. So there's a lot of, a lot of things that people can learn from this. One of the reasons that Legal Riz is so successful is because of the leader of our legal team. When you think about transformation, the legal organization at your company may not be the first group that comes to mind when you're thinking about really rewiring from the inside out. with Agentix, but ours is. And so I want to welcome up the leader, Sebastian Niles, our president and chief legal officer. And Sebastian, we've been able to do these awesome things working together in a rapid innovation squad, but we've improved the compliance, the security, and the trust at the same time, not just made it the same, but improved it. Can you share your thoughts on that?

Yeah, happy to. Look, first, Incredible partnership.

Thanks, man.

It's so important, both for the future of our functions, for the future of our company, and for the future of all the customers and our stakeholders. So I remember this morning, Mark and Jensen were talking about when you think of what we're able to do now that we couldn't do before, it might make every organization more ambitious, right? Make each of us and all of you more ambitious. And I mentioned that, right? Because as you noted, we have a very ambitious vision. for our global legal compliance risk functions. And the vision is, how do these functions be AI native, trust first, and agentic? And why? 2 reasons. One, could we define the best of human and agent collaboration? But also, how do we drive velocity with urgency while protecting the company?

That's right.

Our functions in particular care about that downside risk. And we get hundreds of thousands of requests a year. And a lot of these items, to whom do they go? Which of the legal professionals, which of the teams, which of the compliance teams? And is it possible, though, that that routing and the triaging, is that really what we need of these hundreds of thousands of requests to have legal professionals or any of our incredible humans across the company really focused on? And so this project, this initiative is really transforming. You see the different— the returns here. And how does it enable trust and security? In 2 ways. One, By taking the high-volume repeatable work that needs to scale, right? And using agents, using our platform, using these tools so that it gets done, but in faster, simpler ways. And then ensuring that the high-stakes, right? High-impact areas can really be the ones that our teams focus on. That really enhances, right? Trust and security in terms of where the work is. And then the second one is, What you're building, what we're building, it's governable, right? It's observable, all of those areas. And so together, it enhances our license to operate and license to innovate and lets us go faster. I love that.

And I know we've got a lot more work to do together, but could you describe what you've heard about the way our teams are working as a RIS now and how it's different than before?

Oh, sure, happy to. And there's lots more work to do together. Is it possible that with this project, backlog will go away? Can you imagine if there was no backlog? Mainly because you prioritize all of our work.

Yeah, we love it.

But yeah, what is it like? So look, first, the excitement and energy that our teams bring to this with yours. I think Joe highlighted that people want to collaborate. And that's a huge part of what the Risk Squad does. We're also collectively thinking of problems differently. What is digestible? What's specific? As you said, we're not waiting forever. This initiative, which has had incredible impact, Took weeks.

Weeks.

Right? Not months. And I think the other thing I would highlight is, look, in terms of the changes of the workforce, all of our legal and corporate affairs team members, even the individual contributor, but also up to the manager, they now have this agentic span of control, right? Because they're managing and working with these agents that you're building with us. But in order to do that, we need to be opinionated, right? For example, our new gifts, entertainment, meals app. Who here has to go through in your process, like clearance, for these things. And it just moves really fast and is more simple and enables us to both have great collaboration and drive that velocity while protecting the company. So it's fun, people are happy, and it gets stuff done.

Thank you, sir.

Thank you.

We're going to do more work.

Absolutely. All right, I'm going to wrap it up. And I want to talk about how Salesforce is changing from the inside out. And this is not just about embedding IT inside of our departments. We have a RIS team that's partnering with every BU but 2/3 of our delivery is going to be in Riz squads by the end of the year. You can all do this too. You can follow this example. You heard from Joe how we're enabling everyone to be a builder, and I shared about how we're changing our company from the inside out. Now you're going to get to hear from another opinionated stakeholder around how he is rebuilding his business from the ground up. I got to see it in person in Indianapolis. Adam Johnson, our SVP of SMB Sales, come tell your story.

Thank you so much, Andy. I'm incredibly excited to be here. This is really an unbelievable story. Andy talked about why it's important to have opinionated leaders as a part of this process. So let me tell you why I'm opinionated. I've spent over 2 decades as a seller or sales leader in some capacity, so I understand the mindset of salespeople. I know how they work. I know the things that bother them. I know how they make decisions, and I know the things that get in their way. So I have a lot to bring to the table when it relates to the design and build process of new solutions for them. And Andy did a great job introducing me. I'm the SVP of Small Business Sales and Transformation, but he left out one really important part of my title, which is that I am officially an Agentic Builder. And we're going to— let's go. And we're going to talk about how I earned that addition to my title in just a little bit. But before we do, I want to share a little bit about the journey that we've been on. So last year at Dreamforce, we talked to you about where agents were starting to show up in our sales process. There was an experiment here, a pilot there. And this picture is completely different today. We've got agents embedded in every step of the sales process for our customers, from account research, meeting prep, billing, you name it. There are agents sitting beside our reps, helping them get their work done every day. But I think it's important to also understand where we came from to get here, which is that we were dealing with challenges that most sales organizations deal with. We had disconnected tools. We had over 120 unique processes to get a new customer from lead to accounts receivable. And we had information scattered across dozens, if not hundreds of surfaces that our team needed to be able to do their job effectively on a daily basis. We didn't change this by implementing a new process. We changed this by using the technology that's available now to fundamentally change what's possible for our salespeople. And the first thing that we did was think about what's the critical path? What are the non-negotiables for the way that we want to work? We think of these things as bricks. They're the foundation of everything we do. And these things are managed centrally with input from the business, but managed by IT because they're so important. You need to make sure they're secure, that they're stable, that they're functioning, 'cause it affects everything you do. And on top of this foundation, we start to build the rest of our structure. And I think about these sticks as what I call regional or functional agency. So this is where rapid innovation teams, like the ones Andy talked about, can partner with opinionated leaders like myself, like many of you in the room, to tackle specific problems that unlock productivity gains for teams, departments, or a specific region. And on top of those sticks is where we lay our straws. And this is what I think about as local or individual agency. So how can I, as a person, as an individual in the flow of work, build things that are going to help me at the task level? or with my daily productivity. And it's important to note that none of this is about minimizing the importance of IT. In fact, I think it's quite the opposite. It's about giving everybody the tools that they need to contribute in the areas that they're uniquely qualified to contribute. You know, we talked about, you know, security and compliance and all the things. I have no business in that conversation. But as it relates to things like these sticks and straws, I do have a lot to offer, and I can help carry the load, which frees up time for IT to focus on the things that Only they can do. Now, I'm really excited to share with you an example of one of the things that we built using this rapid innovation model, but before I do, I want to share just a couple observations that I've had along the way. Number one is that you've got to have clarity on what is the problem, what's holding your team back. The second thing you need, you've got to have a good point of view on, you know, what a solution to that problem would look like. And the third thing is you've got to have great business partners like Andy and Joe who are willing to include you as a stakeholder to make decisions as a part of this process. And I think this is really important because what I've learned about the kind of design and build process over the last year of working with them and the team, there are a million small decisions you've got to make to get a new solution to the field. And if you don't have somebody around the table that really understands the user and the way they're going to interact with it and the business problem you're trying to tackle, it's really easy for these these programs to get off course. Now, let's take a look at the Signals application. So the problem we were trying to tackle here is we've got— we've documented over 400 sources of information that sellers need to be able to stay on top of what's going on with their accounts scattered everywhere. We wanted to use headless, to use Data360, to bring all of these things together and deliver it to our sellers in one view where they can focus just on taking action. And so here you can see when I log in as a seller, it's by default grouped by severity. But today I want to take a look at this by type. And when I change this view and scroll down the list, I can see that there is an expiring courtesy contract with Summit Trail Software that doesn't have an associated opportunity. So I want to learn more about what's going on here. When I click into it, because of Data360, because this is running through an LLM layer before it gets to the AE, I've got rich context that tells me exactly what I need to know. to be able to take action. So I'm up to speed on the situation, and I'm ready to engage. So I'm going to click email, and this is going to automatically invoke Slackbot, which is going to draft an email for me as a salesperson that I can send directly from this interface. When I send that email, it's going to get connected through activity capture into the system of record. So I, as a rep, am going to get credit for that activity on the reporting for my manager. It's also going to get retired from this signals view so that I know it's been actioned on, and I can move on to the next thing. Well, guess what? Great news, everybody. I connected with Summit Trail Software and we're going to add some pipeline. And what do I do? I go— thanks for the applause. I, um, I, I go back to this view and there's a button right here. I can add this opportunity, which once again gets tied into the system of record, into the CRM, so it shows up in pipeline, shows up in forecast reports. But I, as a salesperson, never had to leave this view. This is an amazing productivity tool. Our team loves this. Why? Because This is information that was scattered. They spent hours searching for this information on a daily, weekly basis that's now getting pushed directly to them. It's amazing. But it's not just a rep productivity tool. This is also a leadership tool because I, as a sales leader, don't just want to understand what notifications got sent out. I want to understand what work got completed. I want to understand what the business outcome of that work was. And it's important for us to have visibility on that. Notifications on their own are not a business process. And so you might think, oh, visibility, great. You wanna show us a dashboard? Well, we do have dashboards. We also actually can interface with the data via MCP in this interface. But I wanted to exercise some local agency, like I talked about earlier. I wanted to create my own aesthetic for this dashboard. So I'm pleased to unveil to you the Signals Leaderboard. So I know it's pretty amazing, right? But, you know, I'm a huge golf fan, and so I wanted to engage with this data in a way that brought me in that was exciting and fun, but still productive and still meaningful. And so I personally used Claude to create this aesthetic, and it's just for me. It's just the thing that I want. And I think this is really amazing. But we actually used Claude in a few different ways as a part of this build process. And you're seeing here an example of that local agency that I mentioned where I actually changed the aesthetic of the data that we were viewing. We used it as a part of the build process. It's what allowed us to move so fast. We went from kickoff meeting to live in the field with over 3,000 daily active users of this in just a couple months. And we, we used Claude as a QA analyst. So when you roll out big changes, what do you want to know? Are the users using it the way that you expect them to, right? That would have taken us hours upon hours of manual inspection, comparing the CRM evidence to the disposition of the signals. Took me about 15 minutes. Really amazing stuff. And what has all of this transformation added up to? We've delivered hours of productive selling time back to our team. We've got a library of reusable skills that our team can pull off the shelf whenever they need them. We're booking more meetings, we're generating more pipeline, we're closing more business, and we're generating better outcomes for our customers. This is what's possible when you give your team the tools they need to build in the agentic era.

Adam, awesome stuff. Really, really, really good stuff. I, you know, the number on here that I, I just want to talk about, just to put your— wrap your head around it, is this $160 million of PIPE. Like, that's real value to the company. That's quantifiable value. You know, we like to say what reads through in the P&L. That's a number that reads through on the P&L. And how do you think about this? Like, how, from a Like, the sky's the limit. Like, just give us a tiny taste of what you think's coming next, because it's not like we're done. We're not stopping.

No, we're not. I mean, I think the cool thing is we're now getting feedback from the field. We're making this product that I showed you better every single day, but we also have so many more problems that we can tackle. We've got a handful of Riz teams going already on different things, and there's so much more that we can do. I just can't wait to keep building, Joe.

Awesome. Thanks so much, buddy. Appreciate you doing this. Okay, so we saw The redesigning the local process of IT. We saw how we interact with the business. Now I want to talk about how we're going to get this to actually work across our portfolio. How do we bring this to our customers now? And we're going to take a little look at a sizzle right here, and then I'm going to bring up Bernhard to actually talk us through that.

Awesome. I love that video. It represents 2 years of work from our team. Hi, everybody. I'm Bernard Slowy. I'm about 5 and a half years in our customer success team at Salesforce. Before that, I spent 16 years at Microsoft and GitHub, all in service leadership roles. I talk to service leaders every single week in my job, and I hear the same piece of feedback from them that I hear from my own teams. Every service leader hits the same wall. What do I mean by that? We would all love to be proactive in everything we do. We would love to call customers for onboarding, greet them personally, greet them, tell them how great we are, how happy they are to use our products. We would love when customers call us, everybody gets the fastest response time. And we would love when emails come from our company, they don't come from do not reply email addresses. Here's the problem. The maths doesn't work. Every company has a P&L. Every company has a CFO probably telling their service leader, you need to do more with less. That's the reality. Even at Salesforce, we have over 1 million customers. We have over 80,000 employees. If we moved every one of them employees into customer success, we still couldn't do human-led onboarding. And I don't know how Marc would feel about being a support engineer, honestly. We'd have to test that one with him. All right, but it's changing. It's changing because agentic scale is breaking the wall down. It's allowing service leaders to ask a question that, frankly, we've often dreamt of. What if you could scale infinitely? What if you could scale infinitely, build an experience with no constraints? That's what we've been thinking about at Salesforce. We started that journey 2 years ago with Casey, our help agent on our portal. It's reactive support. You go, you ask it a question, it will try answer. If it can't, it will hand you off to a support engineer. But over the past 12 months, we've been scaling that experience across the full lifecycle. Inside Casey, we now have subagents, one for onboarding to help our customers get up and running on Salesforce. We have an adoption agent that helps our customers, recognizes features they're not using, and talks to them about that feature, help them adopt it. We have a renewals agent that helps you go back and forth and do renewals with Salesforce. And as you saw in that video, we've expanded channels. We now have our help agent on voice. You can call 1-800-NOSOFTWARE in the US or Canada, English only, and you can talk to our help agent. It's really, really good. You should go give it a try. It will surprise you. Now, what I want to do today is I want to unpack how we did this so you all can do it too. The thing that really was a game-changing moment for us was 12 months ago, our agent force team stood on stage at Dreamforce and announced AgentScript. We completely rebuilt our help agent on AgentScript, and I want to show you that architecture on how it works. So what you can see here is when a customer comes through to our agent, we have an agent router. That router recognizes the intent from the customer, the job to be done, and then it routes to the relevant subagent. So again, you can see them subagents. We have a support subagent, onboarding, adoption, renewals, all specialized in their job to be done. And whenever one of them needs, they can call on our library of utility subagents. And what's amazing about this architecture, it's modular. When we need to do new things, we just add a new Lego brick. It allows us to scale. The architecture allows us to scale. A good example here is human handoff, right? When the subagent— let's say you talk to our support agent and you want to go through to a support engineer, the support subagent just calls the human handoff subagent. It specializes in that, and it's able to hand them over. This is the architecture that's allowed us to scale. The other thing which has been huge for us— and you heard Mark talk about this earlier today— determinism versus nondeterminism. This is not a whiteboard concept for us. What I'm showing you right now is what it looked like when we handed over to a human in our old architecture. This is the JSON blob that we would send to an LLM for human handoff. It's a lot in here, right? I printed this out and it was 10 pages long. 10 pages. So guess what? We would send that to an LLM. An LLM brings judgment, and sometimes things would go wrong. I would often get tagged on LinkedIn to tell me case creation wasn't working. MVPs would often Slack me and tell me it wasn't working when it went wrong, as they should, right? Because this is a moment of truth. It should just work, and it wasn't working every time. So now that we're on agent script, what you see is we have a logical flow, right? If this, then that. If this, then that. Send to this support engineer. It means we've increased our accuracy, it's more reliable, and we don't need to use tokens when we're creating a case. We can save the tokens for the harder things. This is a great example of probabilistic determinism, right? We save the LLM for the harder stuff. Okay, the thing I'm going to finish with today is probably the thing that I've been most excited about in my 2-year journey on our help agent, and that is launching our help agent inside Salesforce. We started a couple of weeks ago rolling out the agent by replacing the question mark experience in Salesforce. And what I love about this is it has context of who you are, what you're doing, what page you're on, did you get an error message, all of this rich context that we feed into a data graph that powers the agent. You don't need to log in anymore. You're logged in on Salesforce already, right? You don't need to go to a portal. You can do all of this inside Salesforce. Now, what I want to do is I want to show you an actual demo of this. So we're going to switch gears a little bit. I'm going to introduce Emily Winslow, who's our product leader for the help agent. Say hello, Emily.

Hello.

And I'm going to become John Smith. I lead an SMB support team. We're pretty new to Salesforce. I'm inside my service console. I'm looking at my support tickets. And what you'll notice is that I get a pop-up in Salesforce. This pop-up is coming from that same data graph, right? We recognize there's a feature that you haven't turned on yet that would make you more efficient in Salesforce. So I'm going to click on that pop-up. And now what you can see has happened is the help agent is inside Salesforce. I'm not going to a portal anymore. It's coming into my flow of work. We're scaling to where our customers are. And you can see I get this beautiful, rich experience that talks to me about that feature. How do I enable that feature? It greets me personally because it knows who I am. I don't need to log in anymore. And I can go back and forth. I can ask it more questions to help me more on it. So this is it. This is our help agent inside of Salesforce. We're gradually rolling this out so that all customers will have it in our core orgs. by the end of October. So we're really, really excited about that. All right. Yeah, we can get a clap for that one. Go on, that's okay.

All right.

So with that, we're going to go to our last pattern and I'm going to call Joe back up on stage. I'm also going to bring up our incredible People Officer, Natalie. Would you like to come up on stage? We're going to talk about the future workforce we're building. Over to you, Joe.

Awesome. All right. Hello. How are you doing?

Fancy seeing you here.

Yeah, indeed. You know, a year ago, it wasn't actually this stage, but, you know, metaphorically this stage, it was actually over there. We stood here and we were like, wow, things are moving really quick. And then they now move 10 times quicker. So that feels like that was 10 years ago. But I think one of the things that's amazing is obviously all of this is underpinned by people. Like, if you don't have the people resources, you can't make the kinds of transitions we have. So I pulled this together. I just want to show you some data, and I'd love for you to riff on this a little bit about how you're seeing this, but it's pretty incredible what's going on.

It's insane. I mean, we talked even last year, yes, it's a tech transformation. I think many of you were in the keynote this morning, but fundamentally everything that we just saw, even from Adam, from Sebastian, it's a people transformation. And so for us, the biggest, biggest kind of initiative has been that every single job is getting redesigned. And we've had to redesign how we work, of course, in HR, because the stats show it. 70% of work is going to change in the next few years. And 84% of leaders have not redesigned the job. So it's great that we have the technology, the advancements in AI, but if you're not fundamentally redesigning your workflows, it won't be unlocked. So that's what we've been focused on.

Well, you know what, I love this line that Mark used in the keynote is, you know, be the change you seek. And so to your point about redesigning the way that you do the job, you've been redesigning the way you've done your job. And this gets back into the everyone can build.

Yeah, everyone can build. You know, the intelligence that is coming into every employee's hands is abundant. And so now, as you said, everyone can build, and we have seen it as part of our Customer Zero story. So at Salesforce, we're in 50 different countries, almost 90,000 employees, and over the last 8 months, almost 18,000 people are in jobs that didn't exist. So completely newly created jobs. When we think about agentic sales, agentic finance, service, marketing. And that has been huge. And for the HR organization, kind of that more traditional ops role, those folks are now not doing case resolution. They are focused on how do we bring an agentic employee experience into the flow of work to our employees at scale.

Yeah, well, that's awesome. You know, for those of you who've ever seen me do one of these things, I hate slides. even though you've seen a lot of slides, sometimes they're helpful. I like demos.

Yeah, I do too.

Let's take a look at what you built. You built with our teams in this Riz model.

Let's walk through a demo because it's real time all the time, Joe. Okay, so next-gen support. Our biggest kind of— one of our biggest lifecycles in the employee lifecycle is when you are onboarding. We have hundreds of new people every single week, thousands of people a year that onboard with our company. And, you know, the days of logging tickets, trying to figure out where do I find my phone, we're spending way too much time on the bureaucracy, on the grind. So we rethought it and we've completely transformed the experience. So this is Sarah. She is a new hire and she's onboarding. And before she even gets in the building, she needs her phone and she also needs an employment verification letter. And because she's buying a car, so—

Go, Sarah.

Now, go, Sarah. Exactly. So we need to get that to her now. She doesn't have days to think about, how do I get these things? She wants it today in this moment. So Slackbot. Slackbot is the biggest single gift to our employee experience, and our HR operations team have now redesigned their job, reskilled themselves, so that they're training agents to get what Sarah needs in the context that she needs it with the data. So, you can see here, she needs the employment verification letter. Slackbot knows that it's her. They know her permission settings. They know the data and the context, and they're able to provide that in seconds, not days. So, think about that now at scale. All of the time saved to give back to our employees has been a huge agentic unlock Well, that is really amazing.

I thank you. The combination of, again, the business working with IT, small teams, rapid innovation, it's just incredible. So thanks so much.

Yes. Thank you so much, Joe. Good to see you.

Good to see you too. So I want to talk a little bit about, you know, this was sort of an opinionated view. You can see some of the data that's here, but this is part of the Customer Zero story with Salesforce. We don't just solve this for ourselves, we express them in products. So, a lot of what you just saw Natalie do made its way into HR Service in Slack, especially. This is a cool, cool, cool product that we just scratched the surface of. So, this is an example of us trying to solve problems for ourselves, and then us trying to solve problems for you. Now, this is a customer zero demo. However, I couldn't help myself. I had to invite a customer. You know, at Salesforce, we love our customers. We love to celebrate what they've done. But I'm gonna start with this quote. So this is James. He is the CEO of PenFed Credit Union. And so PenFed is kind of what it sounds like, the credit union for the Pentagon. And again, you know, regulated industries, government, all these things, you wouldn't think that they're a bastion of innovation, but they are. And this quote actually was from their earnings call. So we talk about delivering value that reads through in the P&L. They're talking about earnings. But, you know, their CEO is a great guy. He's done a lot of this. He's a visionary. He's driving. But we all know the CIO is where the work gets done. So to do that, I'd like to welcome Sri onto the stage to give us a little bit of a story about this.

Hey, John.

Thanks so much. Really appreciate you coming. So, Sri, you got a tough journey in the sense of you do have a visionary CEO who's pushing, but he has the confidence to push because a lot of what you have done has given him that confidence. So just, you know, give us a little bit of the journey. What did you start to work on? How did that blossom? And how did you get into the point where you can start talking about the stuff that you're doing as a CIO translating to earnings?

Yeah, Joe, thanks. It's incredible to be here. It's actually a lot easier and goes faster than we think. And the secret is always with simplicity. And speed is always the biggest dividend of that simplicity. And here's the backdrop. When our journey began about 5 years ago, our core guiding principle was to strive for simplicity as we transform. And that guiding principle landed us with the world's first unified digital banking platform on Salesforce. This is where you applaud. So we've rationalized more than 400 applications into Salesforce and about 5 or 6 API technologies into MuleSoft. And each phase got faster. Our first phase was transformation of our branches and call centers, and that transformation took about 2 years for us. And then quickly after that, we've transformed our online banking platform on the same tech stack, reusing the same assets that we built, less than 9 months. Next phase of it was a mobile app, and that transformation took less than 6 months. And more recently, our first agent in under 2 weeks. So we now have this unified digital banking platform serving all the products, serving all the channels with over 300 digital assets. And see, when you have data, when your security components, all the assets in a single fabric on the same platform, enabling AI gets easier. It's a matter of enabling those agents within the existing workflows. So that's the biggest benefit with the strategy that we took. And for those who are contemplating, I can give a couple insights. The first is start something internal. And you build your trust, you build your confidence, and you can take that maturity towards external-facing use cases. So the first use case that we dabbled into was IT workforce agent, Agent Penny. contains more than 80% of our IT service desk calls and saving us roughly $2 million a year. A second use case is in the call summarization. PenFed handles roughly 10,000 calls on a daily basis. Agent Wingman, who doesn't like a Top Gun reference?

That name's so good.

Agent Wingman is able to shave off 3 minutes out of every single call and saving us roughly 30,000 minutes every day. So the takeaway here is strive for simplicity. And, you know, complexity is the ultimate tax on speed, on innovation, on security. So when you, when you try to remove that complexity, you get speed and speed gets you the success.

Well, you know, your story is amazing. As Mark said, you know, this morning in the keynote, part of what we want to do here is inspire you. But part of what I'm inspired by is our customers doing amazing work. And thanks so much for coming and sharing a little bit of that story. Thank you. Really incredible stuff. Okay, so we're going to bring it home. We're going to bring it home here, and I just want to review what we talked about. Again, we're building a platform for you that you can use and customize so that everybody in your organization can build things like Claude Force, AI Force, Slackbot, Coworker. These are all things that we're trying to provide to allow you the democratization of everybody can build. We're trying to inspire you with our Riz story about how you redesign the way that IT interacts with the business to go fast in small teams, and deliver value continuously. We're also thinking about how the business stakeholders can really drive more of the vision of what they want much faster in that IT platform engagement model we just talked about. And you saw that from Adam. We also— what we saw from Bernard is like, how do you actually scale this across the customer surfaces? So you all will or have experienced Casey in the Salesforce screens. And that's a critical part of like being where your customer is, meeting them with the right technology where they are. And then with Natalie, we saw how we're thinking about reskilling and the change that we want to see internally to build that future-ready workforce for you, to be able to show you and inspire you as to what we're doing. So that's a lot of information. There's a whole lot more detail too. So I wanted to just point out a couple of things. The first is if you hit this QR code, what you get is how do you scale your agentic enterprise? And so you're going to see a guideline with more detail about what we just talked about. But just as a bonus, just in case you weren't here last year and you didn't get to see the first part, we're also including the guide updated from last year. So if you haven't started your journey yet, you can read that and then read volume 2. It's even better. The other thing that we're doing is we're really trying to get out and meet you where you're at, which is Dreamforce. And so if you look right here and you hit this QR code, you can see we have 100+ that are just the customer zero stories. And what we've done is we've kind of broken these things down that we've shown you and our journey into like modular things where we can go much deeper. So there's over 60 theater sessions, there's demo booths. Go down to the campground if you haven't been there. First off, it's just cool. We built like a forest and all sorts of crazy stuff inside. It's really something to behold. But you can go and talk to the practitioners. So I've brought some of our actual IT team out to sit there and talk about it. You can lay hands on keyboards and experience some of this yourself. So definitely do that. And then we also have the learning labs. So Mark talked about Cloudforce and things like that. There's Customer Zero DNA waved into all of that. So with that, I just want to say thank you so much for your time. Thank you for having confidence in us. And if you have a second, you know, give us a little bit of feedback so we can be better for you. Thank you all. Enjoy the rest of Dreamforce.