TDX 2026 Main Keynote | Replay

Please welcome President and Chief Marketing Officer, Salesforce, Patrick Stokes.

All right. Good morning. Good morning, San Francisco. Good morning to everybody joining us from around the world, to our customers, our partners, our employees, of course, our incredible Trailblazers. Good morning. Welcome to TDX. We are so happy to have you all here with us today. You know, this is our annual developer conference. Where we bring together all of our builders, our developers, admins, architects, everybody building on and with Salesforce, and we walk you through all of our latest innovation, all of the tools and capabilities that we've built specifically for all of you, everyone here in this audience building with Salesforce. And I don't think there's been a more important time to have this conference, to have this event, 'cause I don't know if you've noticed, but the way that we build software is changing like a lot. The way we built software 18 months ago is different than 12 months ago, is different than 6 months ago, is different than like yesterday. It is changing so rapidly. So, we want to spend this next hour or so unpacking all of the new tools and capabilities that we've brought together for you. Now, of course, before we get started, we have to do one thing, which is of course we have to thank you. Thank you to all of our developers, our admins, architects, customers, employees, everybody joining us from around the around the world watching us online. We appreciate the time that you spend with us, all the feedback that you give us, all of the challenge that you give us. We really appreciate it, and we really couldn't do it without you. Now, I don't know if you know this. I didn't actually fully realize it until just a few weeks ago, but this is no ordinary TDX. This is actually a very special TDX because there's a birthday in the room today. And that birthday is not any one person. It's actually TDX itself. Today is the 10th anniversary of this event, of TDX. And really, you know, the timing doesn't quite work out, but it's close enough. And what that really means is that this is the 10th anniversary of the entire Trailblazer movement in general, which is just amazing to think back. Now, we spent some time, we started pulling some photos from the archives, and we were thinking, what can we do to celebrate this moment. And then it hit us. We had an idea. There was one person in particular that had the vision for this, that had this idea to bring all of these people together, all of you that are building and help create an identity and a sense of fun and excitement in everything that you do. Now, that person, she's been off building her own company for a little while, and we're really proud of the work that she's doing. But we thought, what if we could get her back just for one day, on a Salesforce stage. So I'm so excited to welcome back to the Salesforce crowd for a minute, Sarah Franklin. Sarah, welcome. Sarah, you've got some fans. You've got a few fans.

I'm going to cry. Thank you, Patrick, and thank you, everyone. And just For the record, it's not just me. This is a vision of many people that had the belief in what this community could and should be. And it's incredible to think it's been 10 incredible years together. And I'm so thankful to be here with y'all today to celebrate.

Well, I think you've had an incredible impact on so many people in this room, myself included. A lot of people don't know this. You pulled me into the platform team, and then you went to marketing, and you pulled me into marketing as well. So it's just been— you've had an incredible impact on my career, and I think so many people here. But just, I thought maybe— I thought maybe just for a moment, I think it's really important that we just kind of go back, right? And just reflect on why did we do this? Why did we create— you know, you had this vision. You know, we've got dancing mascots, right? Like, how did you come up with this? And why is it so important that everybody, you know, has this Trailblazer mentality?

First, you forgot the bedazzled shoes. We have bedazzled shoes in the house. We have bedazzled bracelets. Like, we bring the shine, we bring the smart, but we bring the sparkle. Of course, you see with our golden hoodies and everything. So we brought the energy.

Yeah.

And you asked the question, why? Well, you know, a decade ago we had this conversation with Marc and with Parker and with many about the incredible opportunity that Salesforce brought, not just for companies, but for communities and for people, for people to blaze trails and do incredible things. You didn't have to have a college degree. You didn't have to have a computer science degree to earn six-figure salaries in US dollars. Having a career in tech. And we made that pathway for people and we said, okay, it's not just about you, it's about the people behind you, around you, above you, below you. You bring them all together. And that's why we did it, because we believe in a world where there's opportunity for all. This moment, AI is actually an incredible moment right now because we're all on the same starting line. Tech has had traditionally a problem with not having equal representation. And this is an opportunity for us. So that's what it means to blaze a trail. That what it means to be together here with Salesforce is that we care about the people first. At Lattice, is what we do is we care about people and their performance. And as much as we'll see about AI and the agentic enterprise, the most important thing is that we do it for people.

Yeah, absolutely. Absolutely.

Well, thank you.

Thank you for regrounding us and our people. We're going to hear a lot about agents today, but we're going to hear just as much about all the people that are bringing this all together. You know, we were backstage just a moment ago, and you were showing me a jacket that you had, and you said, should I wear the jacket? And I said no. And that was not fashion advice. That was practical advice, because believe it or not, you have not yet joined the ranks of our Golden Hoodie winners. And we thought that we would change that today.

So here you are. I thought you would not wear it. Uh-huh. You're going to put it on her all of a sudden. I have never put on one of these. Never? Now you are. Look at that.

You look amazing.

You look amazing. Let's get a photo.

All right, Sarah, thank you so much. I'll give this to you. Enjoy the rest of the show. All right, the OG, the mother of TDX, Sarah Franklin. Let's give it up one more time. All right, so let's get into it. So of course, we need to start the show the same way that we always do, which is of course reflecting on our values, our values of trust and customer success, innovation, equality, and sustainability. You know, these values, they're more than just words on a slide. Right now, they literally are just words on a slide, but they are so much more than that. It's baked into the way that we operate this business, the way that we build our software, and you can really see that when you get into the way that we give back, idea that we call the 1-1-1 model, this idea that business can be the greatest platform for change, where we give back 1% of our time, our equity, our product back into the local community. And I say this every time I look at this slide, the thing that impresses me most is not Salesforce's impact, but it's the impact of the 19,000 organizations, the people in this room that have joined and adopted this model as well, and I hope 19,000 more will do it over the next few years. Now, you might wonder, where is all of Salesforce's impact going? And in many cases, not entirely, it's going all around the world, but in many cases, it's going right here into our own backyard, right here into the Bay Area, into San Francisco, into Oakland, and in many cases, it's going back into children's health and into local education, local schools. This is where I like to spend my time with the kids, showing them how these new AI tools work, because look, if we're all gonna lose our jobs to people that are better at AI than us, it might as well be to our own kids, right? I think that's a good way to get after it, yeah. So thank you. So look, running an organization, a business of this size, while giving back this much, it can be challenging to do, but we do it. We believe that we can do well while we're doing good. And you can see this, the results of this, right here on this slide, this incredible growth trajectory that Salesforce has been on. But I wanna make one thing absolutely clear, which is that we can't do that without all of you, the people in this room that blaze trails, push your organizations to connect with your customers in whole new ways. It's because of all of you that we've been able to grow, and it's because of all of you that we've been innovating for 27 years, building new capabilities and new tools, bringing people into the era of the cloud and into SaaS, bringing people into the era of social and mobile, into the first wave of AI, and now into this second wave of AI, this wave of agentic AI. Now, this is, I think, in my opinion, the most important technology shift, disruption, change, whatever you wanna call it, that any of us will experience in our lifetime. And it creates an amazing opportunity, an opportunity for every organization to change the way that they build, the way that they operate. And we call this opportunity the agentic enterprise. And we think that every organization, every business from small to huge, need to become an agentic enterprise, need to move to a model where humans and AI agents are working together every workflow to build better customer experiences, to build, uh, processes to deliver productivity at a rate that we've never seen before. And for many of you in the room, you, many of you are coders. You can see this, you feel this that first time you start using coding agents where you start to realize and start to feel and see that maybe I'm not one engineer anymore. Maybe now I'm an engineering manager. Maybe I'm really a product builder at large and I have a team of 100 coding agents who can help me build. This is an incredibly powerful motion to go through, and we think every organization needs to do it, and many are already getting started. We have some incredible customers that are already becoming agentic enterprises. Take Lumen, one of the country's biggest telcos. Lumen, you know, one of the problems with a telco is your customers, you need them to renew their contracts when they're up. This is a time-consuming effort, and sometimes things fall under the cracks. So Lumen turned to AgentForce. They're using AgentForce to become an agentic enterprise and now saving over 300 hours per week and picking up, making sure all of those contract renewals are happening. Or take ADP, one of the world's largest HR service providers. Did you know that 1 in 6 employees in the United States is processed by ADP? That's just an incredible number. ADP fans over here.

All right. Great.

I lost my train of thought. You know, you expect applause moments and then you don't expect to play. Anyway, so here we go. So this is, you know, when you— HR, this is a tough environment. You know, people have questions about their paycheck, they have questions about their onboarding, and they get a lot of calls. And ADP turned to Agentforce to help them there as well. Or maybe PenFed. You know, PenFed, this is a credit union.

This is a—

regulated industry. This is a tough place to be. You have to get things right because there are rules, regulations, there are laws. Worst, you know, best case scenario, you break it and you get fined. Worst case scenario, you go to jail. And so PenFed turned to, turned to Agentforce for IT service. This is new from Salesforce. You know, we all have these devices, phones, laptops, software. I have a clicker. You know, sometimes these things don't always work and you got to go create a ticket. Right, inside of it. Well, now AgentForce can handle those tickets for you as well, and PenFed is handling 75% of their routine tasks. But they're not the only ones. We don't just want our customers to be out there doing this themself. We wanna do it internally, because you know what? It turns out building these things, it's a lot different. It's a little tricky. You're gonna learn a lot today about the difference of building deterministic systems, where you have an input, and you have an output, and you can always count on that output. In a probabilistic world, you can't count on it being the same, and so we wanted to go and do build with these new agentic tools ourselves. And we've done that. This idea we call Customer Zero. We're now handling our service cases on help.salesforce.com. Every single developer inside of Salesforce now is coding with agents so we can innovate faster than ever. We are learning what it takes to do this. And I think every organization should be doing this, certainly with, with developers, but also with this help.salesforce.com idea. This idea that now we can have agents helping our customers quickly. This is a clear, use case that everybody can go get started on right now. Now, we don't have it perfect yet. Some of you over here have told us that we don't have it perfect yet, and we're working on it. And you're going to meet the gentleman in a moment who's working on it with us. But we are excited that just a few weeks ago, or maybe last week, we launched AgentForce Voice for 1-800-NO-SOFTWARE. This is our 1-800 help number that we've had for 27 years. It is now powered by AgentForce. I highly recommend all of you go call this when you get out of this. Don't do it now, please, but call this number when you get out of the room talk to Agentforce, I think you will be blown away by the experience. And if you're not, just tell them I sent you. Actually, don't do that. Tell them Joe sent you. And if you don't know who Joe is, I am about to introduce him. So, to walk us through all of this innovation and to walk us through how Salesforce specifically is becoming an agentic enterprise and how to use these tools, we want to introduce someone new to the TDX stage. He joined Salesforce about a year ago as our head of IT, our CIO, and we've recently giving him some product as well, quite a bit of important product. So please welcome our new president of Enterprise AI, Joe Inzirello.

Joe.

Appreciate it. All right. Good morning, TDX. Thanks a lot, Patrick. I'm going to take a little bit of a plug because I'll tell you about the 1-800-Software-is-awesome, the voice, but also the slide before that when we talked about the help site. Just a shameless plug. I did that in my last gig at SiriusXM. I built a help site. I did it. It was underway, but I really leaned into it. Here at Salesforce. Everybody needs one. Help.company.whatever you have, fill in your name there, needs one of these things. It's the best use case because it's really straightforward. You get the ROI right away. I think it's fantastic. So just a shameless plug because I think it's one of my favorite use cases with Agentforce. But here we're going to talk to you about a lot of other agentic stuff. And I think, you know, Patrick mentioned this at the beginning, and I like that integration graph that we had that shows all the different stuff that we've been doing over time. And the thing that's really interesting about it is there's a separate graph, you know, separate from mobile, separate from these other things. It's the nature of the way that we develop software. And so when I was in college, I learned waterfall. Waterfall was sort of a lot of planning, a lot of really refined requirements, knew what you were going to do a year out, sometimes 3 years out. But waterfall was one of those things that was very popular. But the advent of the internet, the advent of the web, really drove a change. And I think the speed at which people were moving doing, the speed at which you could deploy software was very different. And so we moved to agile. And that's really kind of ruled the day for the last 20-some-odd years is this agile development, right? Points, rapid innovation, you could deploy right to the web, you didn't have to get to distribution, you didn't have to shrink-wrap the software in a box. But that transition took a long time. Took many, many decades for that to really cut over. This one that we're in right now, the agentic revolution, is by far the most impactful thing. That I've ever seen in my entire life in computer science. And that agentic revolution is interesting for sort of two reasons. And I think it's important to talk about both of them. The first reason is that software traditionally, with some exceptions of some AI models and things like that, that we're running in the background, has been deterministic. You put in a set of data, you write some procedures, and the same thing comes out the same side every day. Every day exactly the same. 1 1 2. And so we're writing that now using AgentX. So we have these amazingly powerful coding agents, and you're going to see so much stuff that we're doing with those coding agents. It's going to be incredible. But we're also using those coding agents to do a totally different type of software development, which is the probabilistic side. So when you think about an agent, when you think about generative AI, it's not going to do the same thing every time. It's kind of a little bit like a person. Even if you're a subject matter expert, if you get asked the same question on repetitive days, you will not answer it exactly the same way. And while these are not human, they are digital labor, and that software development lifecycle is really different. And so we're going to talk a lot about that coming up. But getting that kind of grounding of the fact that we are now building both these deterministic things and these agentic things using agentic tools is a very, very different way of doing it. So what does that mean for the builders? So you all in this room are builders, you know, whether you're a developer or an architect or an admin. What does this mean for you? The thing that it means for you is you have never been more powerful. These tools are absolutely incredible. It's amazing. Anybody who's used them, there's a before and after moment. There's a, oh yeah, I used to write code in this IDE, and then it's now I talk I talk to an agent and that agent writes the code for me. It's incredibly, incredibly impactful. But there's something that's a little off. And I think especially when we think about the enterprise use cases, why are a lot of enterprises struggling to actually get these agents deployed, to take advantage of this agentic revolution, as powerful, as incredible as it is? Why? I'll think, you know, I'll give you a little hint. 55% of companies, enterprise companies, say that the biggest problem that they have is they do not have enough AI talent. So they want to do it. In many cases, their board is telling them to do it, but they just don't have the kind of depth of talent. Okay, that's where you all come in. So despite what you may have done with Salesforce, in the past, despite the fact that you probably spent most of your careers in determinism. Today, we are going to show you how that changes. Today, we are going to show you how you can use Salesforce Headless 360 to change the way that you're fundamentally developing code. Not just the agentic side, which we've talked about at TDX and other things like that, or at Dreamforce and other things like that, but also the deterministic side. So how can you really take advantage of this revolution? And this is one of the things that's foundational to it, which is this drawing, which you've seen probably quite a bit. The system of context, where the data lives, how does it define and ground your organization? Absolutely incredible wealth of data in here that's required to build these experiences. And then you have the system of work where a lot of you Trailblazers probably spend a lot of time configuring, using sales, service, Sales, marketing. Now those things are all integrated into that layer of work with all of the cool acquisitions and things we've done along the way. And then the system of agency. You've heard a lot about Agentforce. You're gonna hear a lot more about Agentforce. Slackbot, other cool, interesting things that are agents working for you to help you do things. And then the system of engagement. It doesn't take a long time to rewind back and say that really the sales force system of engagement was Lightning. But now it's much richer, much more extended. All these surfaces that you can actually do things with, like Slack, like React, like some of these other agents themselves, being inside some of these agents. And we're going to show you that. But while this drawing is cool and incredibly powerful and grounds everything that we're going to show you today, this slide I think is more exciting. Because at my heart, I'm a nerd, and nerds like REST APIs. They also like MCP APIs, as it turns out. And so every one of these systems is represented in these endpoints. And these endpoints are what you're going to use, or more accurately, the agents that are working on your behalf are going to use to build both deterministic and probabilistic experiences with agents. And this is incredibly, incredibly powerful, this decomposed system that you can compose into things that you want and directed agentially. But not only are we giving you all these endpoints, we're giving you an opinionated view of how you make them work in concert with each other. So one of the things is, if you think about a deterministic workflow, these are kind of the steps that almost everybody does, right? You build, And you can see a bunch of things up there. You could take a picture of it. We're gonna show you a whole bunch of it. So, you know, there'll be money, plenty more time where we dig into some of these things, but we build using these set of tools, agentically now. We test and evaluate agentically now. We deploy those things that we have built and then we observe them to make sure that they're working. That's a pretty standard software development pattern. It's a pattern that we've been doing for a long time. What's changed about it is not just the tools that allow that to happen, but the fact that you can scale this process up and you can actually add the components that are required that are a little different when you're talking about probabilistic agents. Experimentation is a huge one. These agents are very complex. It's not always obvious that if you make one change that you're not going to have an unintended result. In the way that the agent performs. That is crucial. And we're going to go deep into that today. The control and orchestration, like a single agent is amazing. A legion of agents changes the way that your company works. And that is the reason that we're going to talk to you a lot about control and orchestration as well, because that is a huge unlock. That is the thing that really allows us to scale beyond just sort of the individual contributor world to this massive— the whole company runs differently. But we're not just doing this ourselves. We're not just doing it for ourselves with our own tools. We are absolutely 100% committed into working on this in an ecosystem that is open. I think one of the things about agents is they hate boundaries. Agents do not like to stay in a silo. They want to talk to each other. They want to talk to endpoints. And that's why it's absolutely essential that you can take a look at this chart and you see the who's who, the hyperscalers, the AI natives, a lot of folks that you probably already work with from a consulting standpoint, all using the same family of tools, this huge open platform that's now there for your taking. And Salesforce is an integral and driving force behind this open ecosystem. And I think that becomes incredibly powerful. The other thing that's incredibly powerful is the Salesforce ecosystem. So many of you have probably bought things from our partners inside of our exchanges of various sorts. But one of the things that we're announcing now is the Agent Exchange. This isn't just full agents, it's skills and other things that you can use in this decomposed composable system that is Salesforce Headless 360. You can use these things to build build your own experiences. So as an individual trailblazer, agent blazer, developer, you don't have to build everything yourself. You can take advantage of the rich ecosystem that we have here and put those into your products so you can move faster, focus on the things that deliver value for your company. Now, the other part of that is— I'm going to talk to the other side of the room— for those of you who want to put things in the exchange, we're announcing this $50 million fund to help you do that. So for— yeah, there we go. So for all the agent blazers, for all the trailblazers out there, you can now build these things and you have a way of monetizing them in our marketplace. And this is just going to add to the hundreds of thousands of people that have donated their time and put product in there that they've been able to be compensated for. And that's an incredibly powerful thing. And we're leaning in. So we talked a lot about the general gist. There's all these sort of things, and we're going to keep going high and low because agents kind of do that, right? They get down in the dirt and then they go all the way up to the high-level stuff. You know, you talk to them in these high-level ways, but composability in that ecosystem is a theme I'm going to keep coming back to. So I want to introduce you to a company that I think is absolutely super fun. Their name is Enjin. For those of you I guess they're not sitting with the ADP people. But you'll know them after this. There we go. You'll know them after this presentation here. But actually, everything we're going to do in the rest of the keynote is going to revolve around Enjin and show you how Enjin did what they did. But they are an agentic-first B2B travel company. So think of them like Concur. But they basically make these APIs available to other folks to dip into it. But they're 100% agentic themselves. And it's an amazing, amazing story. So to do a better job of explaining it than me, let's actually hear from the Enjin team. Enjin was a company going through explosive growth.

There's now 1,000 of us inside the company. It's just scaled tremendously. We're a modern travel and spend management platform. It's a very big space. Very ripe for disruption.

Members of our support team were doing 300 cancellations a day.

We were all doing the best we could with the tools and resources we had available to us at the time.

But there was still only so far you could go with that.

You simply need something that is ubiquitous, that everything can tie into. Salesforce is the operating system that makes it seamless. It's the platform that the world runs on. We had the use case, we had the right partner, Demetra and I just had the willpower. We're going to get this thing done.

We built what's now called Ava, Engin's virtual assistant, in 12 days.

The first few days were spent just opening up Agentforce Builder, typing in prompts.

We tried to break it every which way we could before we put it in front of customers.

I've worked with chatbots in the past, and agents are just a lot different. Trailhead makes it very easy to start learning, and that allowed us to develop Ava very quickly. We track how many of our cases are actually handled by Ava, and we are at 50%. She could now recall information about a booking, answer people's questions about their reservations, saving the reps so much time so that they can focus on things that really matter.

Now we have 5 agents helping across the company, whether that's customer-facing or employee-facing. Agentic AI is only as powerful as the systems that you connect it to. Making sure it's rooted and connected to the right data is mission-critical.

Data360 is at the heart of our infrastructure. Without it, we wouldn't be unified.

And for customers, you're getting what you want faster. Slackbot now is essentially an extension of the knowledge base. New people can go into Slackbot and say, "Who does this?" and "Where would I find that?" Having our entire knowledge base in Slack in a moment's notice is a game changer.

Slackbot truly is the ultimate teammate.

5 years ago, the question always was, "Is this possible? Can we do that?" And now it's like, "Of course we can." Having Agentforce, Data360, Tableau, and Slack integrated in one system It's just extremely efficient.

It's no longer humans versus agents. It's humans with agents.

CSAT goes up, costs to deliver go down, customers are happier. We're getting them answers faster. What's the trade-off? There's no trade-off.

You can feel it from every person at Engin. We all actually believe in this agentic enterprise.

There's just incredible opportunity for all these businesses. If you run on Salesforce today, I mean, you're nuts if you don't.

All right. So, hopefully, y'all have a better understanding of NGINX. I think they summed it up the best. This is built entirely— they're all in on Salesforce, on Slack, on all of our products, which actually just makes them really smart. But in addition to being really smart, they are also participating in this open ecosystem. So, everything that you're going to see from here on for the next 3 demos that we're going to go into, is using Engine as part of that ecosystem. The fact that they're built on Salesforce isn't actually that important. Again, other than the fact that they're smart and everybody should use Salesforce, but it doesn't matter technologically because they are part of the open ecosystem. So right now, I'm going to bring up a buddy of mine, Kush, who's going to tell you how we use these types of things, including Engine, to build deterministic experiences where you're coding with an agent. So Kush, come on up, show the team. All right.

All right, thank you, Joe. All right, good morning, friends. It's so good to see all of you today. Now, what you see up here should feel very familiar. It's our application lifecycle. Design, build, test, deploy, observe. What's different, however, is how we move through it. With Agentforce Vibes, you're working alongside agents that understand your organization's metadata. Data, business processes, helping you accelerate across each one of these steps in a unified, connected, single flow. And with governance built into every step along the way, you can go fast without compromising on reliability or security. Whether you're trying to experiment or whether you're trying to build production-grade applications, you don't have to choose. You get speed and trust from the get-go. And to see this in action, let's see this from the lens of Engine and through a demo. And in this demo, you'll see us touch on a number of these capabilities. And in this demo, we will be jamming with Mr. Dan Fernandez, the product lead for Agentforce Vibe. How about a round of applause for Mr. Dan Fernandez? Excellent. So Engine wants to build a partner service app. And the team, as you can see over here, they live in Slack. This is their project channel where you can see there's a ton of important information about the project that they're working on, this particular service app that they're building. And with Slackbot, we can easily synthesize all of the information, the discussion, the content, the API contracts, the data model designs, et cetera, UX wireframes into a comprehensive product requirements document, a PRD, which we are now going to ground our agents with to begin building in Agentforce VIBES. Dan, where do we start?

So we're inside of Agentforce VIBES 2.0, and the first thing we're going to do, we've downloaded the linear data using the linear headless MCP. And what we're going to do is just run this. This is going to use the Salesforce Catalog MCP. It's going to read the attached PRD and see what data we actually have within the org.

I see. So this is going to initiate the Metadata Grounding MCP. Now, it's our first headless MCP, which allows you— which behind the scenes actually triggers what we call the Salesforce Metadata Catalog. Now, think of the Salesforce Metadata Catalog as a semantic search index on engines metadata exposed to the agent via MCP. Now, this is super critical for those of you who have invested so much into your organization's orgs, your orgs, right? You've put so much into there. You've got so much metadata. You want to be able to reuse all of that. You want to be able to extend that in a governed and trusted manner. And so this really helps you build with context. Dan?

So under settings, let's just show you what's happening under the coverage while this is running. The first thing you'll notice is Agent Forest Vibes 2.0 has this agent harness. And what you see, we have the Anthropic Agent SDK and the OpenAI Agents SDK. Now we're not just using the LLMs, we're using the actual agents. So what does that mean for Agent Forest Vibes 2.0? You get the power of Claude code and OpenAI agents tailored for Salesforce with these headless services all available running securely in the Agentforce Trust Layer. Amaze, amaze, amaze.

Wait, Dan, does that mean that I get all of the various capabilities of each one of these agents?

That's exactly right. So watch this. As I change the agent harness capabilities, the harness capabilities will actually change depending on which agent that I'm using. So you as a developer are able to choose the right agent for your specific task.

Ask.

Excellent. What do you guys think?

Yeah?

But wait, there's more.

All right.

So we have 30+ skills that ship in box. Now, these skills, I'll show one in a second here. We can just filter this and you can see all the different skills. These are just markdown files. This is the agent script skill and this is written and detailed instructions how to write agent script for the platform, what to do, what to always do, and specific tasks. This is how I want you to write this. This is how I want you to change. This is how I want you to validate yourself. And these all ship in box. Now, a couple other areas. We have our model context protocol tools. And again, this includes the third-party tools like Linears, but the Metadata Grounding and all these other tools. We have 60+ MCP commands to make you more productive as a developer.

If I may add over there, Dan, These are MCP tools, these are skills that are all headless. You can access them from any tool of your choice. You want to access them from Codex, you want to access them from Cloud Code, Cursor, WinServ, choose your tool. What we have done is we have packaged them all, we've preconfigured all of them into the Agentforce VIBES IDE. It's available in the Agentforce VIBES extension so that you can get really, you can get started really quickly. But if you want to use all of these in a headless fashion, in any tool of your choice, go for it.

Great. And the last one, one of my favorite features, is the way we would look in developer tools, we'd see an XML file. I think we can do better. So this is a schema viewer that ships. And what this allows me to do is have a rich visualization of the schema. This is custom objects and standard objects. Now for InGen's application, they have over 100 fields. If we were doing this inside of our org, every field takes 7 clicks. We just saved you 700 manual clicks by doing this agentically.

Excellent. How about we go back to our agent chat and see where our agent is? So as you can see over here, it's already determined a set of metadata that exists within Engine's org that we can build upon. And so now that we've got our foundation that we'll be extending, Let's go ahead and build the app. Now what we're gonna do over here is we're gonna ask VIBES to just say, look, go ahead and build this app, go employ a bunch of subagents, build out the app, test it, harden it, deploy it. And obviously it's gonna take a few hours. So what we've done over here is we have fast-forwarded to the end result. As you can see over here, this is the Engine Partner Service app. Now we'll step through it, you can see pretty visually compelling. We've got a fair amount of good data. We've got open cases, you've got your guest satisfaction results, your properties, and you've also got Eva, the agent that was generated. That skill that Dan just walked us through, this was used by the sub-agents, by the agents that we have running in VIBES, to go ahead and build out this entire experience. Now another thing you should also notice is that is that directly within Vibes, you have an inline preview of the application that is being generated. So any changes you make, you'll be able to see it directly in here. Now, there is one thing though. This is not Lightning. This is React. This is the React that all of you know and love. This is the React that AI understands, but natively running on the Salesforce platform. Connected to Engine's org metadata, securely retrieving the data using GraphQL, inheriting all of the security primitives, the access models, the access controls, the full trust layer. This truly changes everything. You know those visually compelling experiences that you wanted to build for your customers, your partners, your employees that you could not do on the Salesforce platform? With Agent4, with skills, with MCPs, all headless, you can now do all of this on the Salesforce platform. How about that? All right, so it's pretty well connected to data, but we want to extend it a little further by connecting it to Data 360 using our brand new headless Data 360 MCPs. Now, the genesis behind this is Engine wants to allow their partners to be able to self-serve and create custom segments. And so what we have to do is we've gotta add a custom segment builder directly within this portal, this app, so that these partners can go ahead and create those segments. Now, what would usually take a few hours, you'll see in a few seconds show up. It's because of these MCPs, these Data 360 MCPs. They give the user, the agent, full control to configure, to manage the Data 360 the data within Data 360. So if you're trying to build any experience using data, these MCPs are your go-to. And so in a matter of a few seconds, you can see we've got the custom segment builder all built and all ready to go. Now, we just walked through a lot in a very short amount of time. There are breakout sessions for each one of these, which we encourage you to go to all across TDX. And so let's bring this home. We just planned our project in Slack. We connected and grounded our agents using the Metadata Grounding Service, the MCP, using the metadata catalog so that our agents know what's going on in the org, what metadata is already there. We used the Open Agent Harness. We used the headless Skills, the MCPs. We used the multi-UI framework capabilities within Vibes, preconfigured all within Vibes to create a visually stunning React application, fully connected, fully governed on the Salesforce platform faster than ever before. Thank you very much for your trust. And back to Joe.

Excellent. Thanks so much, guys. That's fantastic. So that is a great example. Example that we just saw about how you use these agentic tools to create a deterministic experience far in excess and massively faster than you used to be able to do with the Salesforce tools. And I just want to point two things out. One, you saw it behind the scenes, they were using Vibes, but it was Claude code behind the scenes, which also means you could just use Claude code if that's what you're already working in. The second thing is you saw MCP endpoints and we keep coming back to this, I was trying to explain it to a friend of mine and why I was excited about this keynote, and he's not all that technical. And so I said, every time you hear MCP, just think Legos, right? You can build whatever you want. You can compose them in any way you want to do it. And that's a great example of a deterministic experience. If you want to learn more about how to build with coding agents, here's the QR code that you have there. But I want to shift gears because that is awesome. It's, it's real work that you know better than anybody, all of you agent blazers and trailblazers out there, because you do this work every day. So it's exciting. But I think even more exciting is when we get into this notion of building agents. How do we change the business? How do agents step function change the business? And so we're going to talk next about how we scale agents and scale them with trust. You know, I used to think about like when you built a piece of software, 80% was the build, 20% was sort of the refinement. And with an agent, it's the opposite. It's 20% to get the thing out there. To make it great takes the 80% of the effort. And it comes back to this point of determinism versus probabilism, right? This notion that these agents inherently, because of these LLMs, have incredible flexibility, but that flexibility comes at a cost. And that cost is they're not going to do the exact same thing all the time. Now, we've built a ton of stuff into AgentForce to help you control that. That's why AgentScript has determinism in it. It has all of these components to help you with. But you also need a really, really good development cycle. So this sort of continuous improvement loop is such an important part of what we're doing that I really want to go deep on it. And so I got to tell you, it's my absolute pleasure to welcome up to explain this to you, my friend Nancy, who loves A/B experiments just as much as I do. So with that, Nancy, Please show the team.

Thank you, Joe. One of the best parts about my job as VP of Agent Forest is I get to share with all of you, our trailblazers, the newest innovations in AI and agents here at Salesforce. And this year is a very special year because if you are a developer or if you talk to any of your developer friends, you know that agents and AI are transforming software development. And we believe the same is going to be true for agent development. You can't just tune your way and build a prompt to build a good agent nowadays. I think we all figured that part out, right? And you need guardrails, determinism, trust, control. And that is why we built AgentScript and it's GA now. You can use it to write one single file to define your agent deterministically. And we think that's so powerful. Now I'm a developer myself, and I know that we all have the tech stacks that we like to use when we build. And that's why with Agent for Skills that we're releasing at TDX this year, you can use any coding agent of your choice to go from zero all the way to agent in production with skills. And when you have that agent that you love, all you have to do is hop on over to our new testing center experience where we have custom scores, custom evaluators, so you can build those agents knowing that they follow the guardrails for your business. And when you're ready to deploy, it's one click, one agent across all your surfaces. You have voice, Enhanced Chat v2, mobile SDK, and so much more. All these experience layers available for your agent with one interaction. And as Sho mentioned, my favorite part of all of this is that building an agent isn't just about the zero to one anymore. It's about the n to n+1. It's the iteration, the improvements, making your agent better every single step. And with versioning inside your agents and with our new A/B testing API, you can build confidently knowing that your agent is getting better every step of the way. And last but not least, because trust is so foundational to what we do here at Salesforce, we know that as business users, developers, admins, you need to be able to observe, monitor, understand what your agents are doing in production. And with our new SCDM API that is now OTel compatible, as well as telemetry and Slackbot, you can do all of this from the places that you work. And lastly, because we aren't just single agent organizations anymore, you can connect all of your agents inside your org with multi-agent orchestration and with MuleSoft Fabric or Agent Fabric. And these are all GA now. This, my friends, is Headless 360. And I'm so excited to be sharing this with you today. And to make this all real, I want to walk you through a demo of how this all comes together in a real agent build experience. In the first session here, Kush showed you how to build a partner dashboard for Engine. Now, if I'm an Engine partner, my first question is, what happens if I actually want to do something in that dashboard? Right? Maybe I want to escalate an issue. Maybe I want to dispute a transaction. What Engine needs is a partner service agent, and we're going to build that together right now. And to do it, I want to take you to a brand new place inside Agentforce. This is Agentforce Labs. This is my favorite new release at TDX this year, and this is the place where my team and I, we get to ship to you our agent blazers our newest innovations before they go GA. New ways to build, new ways to get started, new ways to test, new ways to build products that actually change how your businesses operate. And inside Labs, we have a special area called Getting Started. And with Getting Started, we're making it possible for you to build an agent and get started with an org in a matter of seconds. And as a builder, this is so exciting for me because I know that sometimes the hardest part about building is just getting all the info you need to actually start getting to the meat of it. So with the getting started page, I can go ahead and type in a description of what I want for the engine partner agent. And in a couple of seconds here, you'll see I can choose from more than a dozen different ways to build. A lot of you in the audience have been asking, how do I use Cloud Code? How do I use Codex? This is it. I pick Cloud Code. I can copy paste the command and I go straight over to terminal that all of us have on our desktops. I have same terminal window we all have. I've activated Cloud Code and I just copy paste that command that I just pulled over. What's happening here? Okay, this is really magical. In one command, I am pulling down all my agents for skills, all my sales for skills. I'm provisioning an org. I'm setting up an agent. Agent script for my agent. And not only that, it's going to deploy it for me too. While this is loading, let me give you a quick preview of how I want to demo this experience to you. First, I want to show you how you can build an agent inside Terminal. We just kicked that off. Second, I want to show you how we can test your agent so you have trust in the performance of your agent. And third, and most importantly, I want to show you how you can improve, observe, understand how your agent is doing so it gets better every step of the way. And we're going to do this all inside the terminal with Cloud Code. And as a bonus, as a developer myself, I want to geek out a little bit with you all about how this all works behind the scenes and show you agent script because I think there's nothing better than actually looking at the code so you know what exactly is happening and you build that intuition for what your agent is doing. And you know, since we're still loading in terminal here, Stefan, why don't we go ahead and just open up agent script straight from our terminal and show folks what this looks like.

Yep, absolutely. Thank you, Nancy. So this is AgentScript. Okay. It truly is the backbone behind AgentForce that makes all of this stuff possible. Okay. What it is, it's a single flat file that's versionable and really controls the way that the agent runs. You'll see here we can store variables so that we can make sure that users are authenticated and sticking the same throughout the session. And then we have the sub-agent selector, which is where the logic really starts. Every time a session starts with your agent, it's going to take in that prompt and determine exactly which subagent it needs to hand the task to. So for example, we have subagents here to manage their portfolios, okay, to go to disputes, to manage disputes, to go to events, and it's able to route those transactions every single time. It adds that level of determinism. Now, for example, one of those subagents here for payout disputes allows you to specify metadata in your Salesforce org like flows, Apex, text classes and bring that back into the agent. But where the real power is, is right here in the reasoning. You'll see here we have really detailed instructions that we need the agent to follow every single time, whether it be this list that it has to go through every time or dependent on specific variables. We can always see here that the agent can flag when it needs to. In this case, if a discrepancy has been found, the agent must always escalate to a case. Otherwise, we don't want the agent to continue and try and process that because that's not what it's for.

Okay?

We have our humans here that we can hand off to and work with inside of Salesforce. And the agent script helps us control that.

This is so amazing. And Stefan, you just talked about two of my favorite things, which is trust and guardrails. I hear some laughter here. So what I want to do is go ahead and type in a quick command here. Inside my terminal. And in order to show you that my agent actually follows the guardrails that I have in place as a company, I can run a smoke test for this agent. And what this is doing is we're going to simulate a real agent user using my agent, and it'll help me understand how my agent performs in these simulated scenarios. And while this loads, I want to show you something else that I like to do as a builder. So Stefan, if you don't mind, let's go ahead and open the preview right here inside the terminal. And I can open up a preview where I, as the builder, can just chat straight with the agent and build intuition on what's happening behind the scenes. This is so powerful because yes, I can run those automated tests, but I as a builder also need to understand myself what's happening. And while I do this, this is manual mode. I can go back to the smoke test and I can see how the smoke test performed. Performed, right? So I've got manual version, skilled version, and I can see that my tests all passed. And this is really cool. But there's one more thing I want to do before we go ahead and deploy this agent, which is I want to take you into the builder. So we're going to go ahead and open up the builder straight from the terminal here. And inside the builder, this is where you can collaborate with your business users, right? I, as a developer, I like to use terminal, but my business users are not so keen on the terminal experience. So we open up the builder for them. And here they have the canvas view, which is the same capabilities you get with script, but inside natural language. You might be asking, well, what's the trade-off? There is no trade-off. You get the same power, the same determinism. It's just a different interface for a different type of user depending on how they want to work. And inside this experience, there's one more thing. There's Agent Exchange. Inside the builder now, you can connect any one of our capabilities across the entire Salesforce ecosystem because of Agent Exchange, and you can bring in subagents right into your core agent, and it's a matter of a click right here. Okay, I'm feeling pretty good. Let's go ahead and deploy this now, Stefan, into Slack, because Engine is a Slack agent, Slack-first company. So if I go ahead now and type deploy to Slack inside my terminal, I can see all of the agents that Engine has built. And one of them is the partner service agent that we just built together here. Of course, Engine has many other agents. And one thing that I really love about how they operate is that Engine has a special channel where they're focused and obsessed about improving their agents for their customers. This is the feedback channel. Feedback channel. Okay, this is really magic, guys. Inside the feedback channel, every agent, every time there is an issue with that agent that users flag, it gets sent to this channel. Their admins, their builders, their developers, they look at these sessions and they help understand what went wrong, how do we do better, and what's the next version of the agent we can ship. And because we have telemetry in Slackbot and because we have this OTel-compatible SCDM, all I have to do is paste that session trace ID and I'm able to analyze is using Slackbot, what exactly happened in the session, how do I make it better? This is improvement in action, one command, copy paste, made it better inside Slack. Okay, as I showed you earlier, we have several different agents all inside the Engine environment, and there's something magical when you can take two agents together and make it 1 1 3. We built the service agent for Engine, Now, Engine also has an SDR agent. What happens if I can bring together my service agent and my sales agent and have them talk to each other? That's the magic scenario here. And so for Engine, every time a partner talks to their support agent, they want to be able to also flag to those partners sales opportunities that those partners should be looking at. And that is what their SDR agent will do. And it's super easy for us to do that. This. All I have to do is then go inside the builder. I click on the SDR agent, I click add, and voilà, that's it. Okay, let's take a step back here. I know there's a lot that we just showed you. Let me just tell you what we did in the last 10 minutes together here. Okay, we just built an engine agent from scratch. We had no org, no agent, and in the last 10 minutes we built an agent, we built an org, we deployed it, we tested it, we observed it, we improved it. All inside Terminal and with observability inside Slack. And not only that, we did this in less than 10 command lines in the Terminal, guys. This is so magical. How cool is that? All right, thank you. And just to show you all this is all real, here is that agent now on the dashboard website that Kush had built with us earlier. Now, this is all really magical. And to wrap up, I just wanna share with you a conversation I had last night at a partner event we had. One of our partners came to me and they asked me, well, hey Nancy, what keeps you up at night when it comes to Agentforce? And I told him, look, the answer's really simple. It's how do we make all of you Agent Blazers in this room, how do we make it possible for you to build agents that are enterprise-grade, trustworthy, quick, quickly, easily, and most importantly, in the tools you love. And that is what we've done today with all the innovations that we've shown you. And my ask is that all you do is go after this keynote, go to labs.agileforce.com and try it out and see some of these innovations for yourself. Thank you so much. It's back to you, Joe.

Wow. I mean, pretty awesome. We saw agents building agents. Deployed to a website that we saw built by agents in an incredibly fast amount of time using MCP. Again, read as Legos from things like Linear and all this stuff in that cocktail. If anybody's having an, like, an Inception-style, like, hallucination right now, that's actually normal. Just breathe through it. You'll be fine because we dumped a lot of stuff at you. But I think if you want to learn more about this particular side, like, how do you build and scale agents, which is obviously one of the things that we think is paramount for all companies to do to become an agentic enterprise, visit the Agent Summit. So now we showed some determinism, we got a website, we showed an agent that's able to take that experience to the next level and represent itself in multiple surfaces. But how do we make it easier for you to deploy on any service? So back to the openness of the platform and the way that we're presenting these Salesforce Headless 360 is the way that you can start to think about this is your gateway to these other experiences and meet your customers where they're at. So on all of these different things, whether it's voice or third parties or Slack or the actual coding agents themselves, this is how you're going to be able to get there. But you don't want to build 50 versions of it. That's the last thing. It's like the browser wars all over again. Remember the days where we had to test in like 10 browsers? We don't want that. We want to make it easier. And so to do that, I'm going to bring up a person that's going to tell you about the Agentforce Experience Layer, which is incredibly exciting, and it's going to allow you to do this at speed. Johnny, please take it away.

Thanks, Joe, and hello, TDX. All right, we're going to kick off this chapter by unpacking how Agentforce Experience Layer actually works. So it all starts when your customers start interacting with your agents. And it doesn't matter what surface, right? Whether it is the websites, the mobile app, Slack, Teams, or any of the third-party LLM surfaces, ChatGPT, Claude, you name it. Now the agent has packed up a response, and all of this is backed up by the MCP service, the Legos, as Joe pointed out. And this basically is a standardized connective tissue that enables your agent to securely pull data from your systems and trigger actions across your entire tech stack, such as update a CRM record or help with a car rental information. Now is when the Agentforce Experience layer steps in. Your agent has a response, and the layer then determines if it needs to render it as simple text or if it's a rich interface. Now, if it's the latter, then it composes the UI, binds the data, wires the action, and renders the interface on any one of those experiences and surfaces that you see up there. So that is a little bit about the end-to-end. Now let's see how all of this comes together with the help of a demo. Let's give it up for Stefan once again, you guys. All right. So we're going to start with the company's website. This is the home page, and on the bottom right-hand side corner, you see the current implementation of the agent. Now it works, it's great, but we're going to make it better. We're going to make it more interactive and more actionable. So as you can see here, a traveler is on the agent and having a conversation, asking for an update about an upcoming trip, but the response is not It is rich interface that is embedded right here in the chat window. This is enhanced chat v2. So what used to be frontend development and backend development and everything in between and orchestration is now collapsed into a single runtime agent. How cool is that? So let's take a look at how this works. We're gonna pop over to our VS Code project. All right, now under the hood, nothing fancy really. This is, is your standard Lightning component that all of you are very, very familiar with. All right, so, so far in the demo, you have seen that whether we start with voice or we end up in the web, you're able to deliver the same seamless conversational and intuitive experience no matter what surface. Now let's get to the meat of the demo. We're gonna show you the Agentforce experience layer. But before that, Think about who InGen's customers are. They are business travelers, right? And by definition, a business traveler is not locked into a single tab. They are not simply on the web or on mobile. They are on Slack, on Teams, and increasingly they are starting their journeys in consumer AI assistant surfaces such as ChatGPT and Claude. So how do you as a developer make sure that you expose this rich experience you have built and make it look and feel native across each one of those surfaces. And that is what Agentforce Experience Layer helps you with. So think about it as a cross-platform abstraction layer for your UI where you simply describe your, your experience, the schema, the layout, the actions, and it generates it for you. As simple as that. All right. So it is actually as magical as it sounds. So let's go take a look at the experience layer playground. On the left, you see all the components and layouts that is backed by that single standard structure on the right. Now, the key here is you just have to define it once and it renders seamlessly across multiple platforms. And in this view, you get to see the code that is backing the components. But let's flip over and see some of the components that we have already built. So as you see, there is a DFW flight status component. We have a seat upgrade component all built up. So how do we build it? Two options. One is you can use agents to generate this right inside Agentforce. And secondly, if you're a visual person like I am, you can go into our low-code builder and drag and drop the components and create that rich interactive experience right here in the Agentforce experience layer. Now we also have a nifty, neat little feature in preview where you can see how those components render across platforms such as Slackbot or ChatGPT or Agentforce. All right, so, so far we have defined and built the experience in the layer. Now let's check out how this looks live in production, starting with Slack. Now a lot of InGen's customers are on Slack for work, right? Remember our agent Ava? She didn't respond. There, but let's see how she does here. Now we have plugged her into the Slackbot interface, and what that means is your users do not need to leave the flow of work. This is a big shift, you guys. This is the big so what. In the entire demo that you've seen so far, you are actually not pulling your customers into your mobile app or website anymore. You're You're pushing your experience and brand and meeting them wherever they are, and that is the power of Headless 360 and the experience layer. All right, so let's take a look at what's going on behind the scenes in Slackbot. Behind the scenes, the agent force experience layer is tapping onto the agent, and then that enables the Slackbot client to construct that rich interactive experience. And what's cool about this is that it is not restricted to a single app or a specific deployment. Even if the agent is not native to the surface, you're able to plug that into any Slackbot instance as you can see here. All right, now we know that InGen has over 20,000 customers, and not all of them are on Slack. We'd love for them to, but we know that's not the reality. Now some of them are on other collaborative surfaces including Teams. So our service agent Ava is now plugged into the Teams surface, as you can see up there. So what this means is that the experience layer, as well as agent force, is running natively on the surface, on Teams surface, so you can then provide that same interactive, rich experience to your customers wherever they are. And likewise on any mobile app as well. With the help of Agentforce Mobile SDK, you can now take that experience that you've defined and make it a first-class UI right here inside of the app. So as you can see, when a user hits a push notification, they land right into the rich interactive experience that they're used to seeing across platforms. All right, so, so far we have seen how InGen reaches their customers on their everyday workspaces, right? Like Slack, Teams, mobile app, et cetera. Now this time around, we have made ChatGPT the target for the service agent Ava. So we are bringing agent Ava right into the context of the ChatGPT interface. So she's part of the conversation, and what this means is now the experience layer is rendering it it and adapting it to the context of the ChatGPT instance and the surface. And while we're showing you this on ChatGPT, this can easily extend across Claude or Gemini or any other surface as well. Now, while we're here, I wanna show you one last thing. Now, what if we wanna go beyond simple queries? What if we want to show an account plan for one of InGen's customers? Now, as you see there, Agentforce for Sales is embedded directly into ChatGPT, and behind the scenes, ChatGPT can reach out to Agentforce in real time and bring back dynamic account details. It's not static, so it's part of the conversation. And now you get to ask follow-up questions and drill down for further insights as well. All right, I'm going to take a breath here. We have shown you a lot of things. So let's pause and connect the dots across all the demos you have seen so far. Now, we started by defining the experience once in the experience layer, but then we deployed it on not one, not two, but six different UI surfaces, six different platforms. Now, a little personal anecdote. I am also a recovering full-stack developer. I started my career 18 years ago as a software engineer, and I just realized that I may have dated myself in that process. Taking one for the team, always. Anyway, back to— if I wear my developer hat, I am handling 6 different UI surfaces. I'm managing state, auth, edge cases, not to mention version drift. Everything across 6 different platforms. And you guys know the complexity more than I do. This is easily months, if not quarters worth of effort for a team of developers across different surfaces. Now, what we just saw is with the power of Headless 360 and the experience layer, we went from a PRD to production to in the hands of your customers across multiple surfaces. In just a matter of weeks. This is not incremental. This is 5x, 10x productivity gains. So I want to leave you guys with this one last thing. Wherever your customers are and no matter how fast this ecosystem is changing, you are able to meet them wherever they are faster than ever before with Headless 360. And with that, back to you, Joe.

Great. Thanks so much, Johnny. All right. Really slick. So I think, you know, deploy on any surface is really powerful, but I want to ladder it up. But before I do that, like, take a look at this. If you hit the QR code, you want to learn more about that. There's a lot more to learn. Like, we went at light speed to get this thing done. Okay, now I want to bring it back. We used an agent to build a deterministic website very quickly using multiple tools, not all of which was Salesforce. Participate in this open ecosystem to get that built. We then added an Agentforce agent to that website, and we grounded it in the Salesforce organization that existed there, and we gave it great capabilities. We then broke it out of the website and made it appear everywhere that your customers are actually in the flow of work. And we did all that without you having to write it. Now, I think you're probably thinking yourself, this is a very strange Salesforce presentation because you didn't You didn't see anybody write Apex by hand. You didn't see a whole lot of Lightning interface. But what you did see was the future of what you're going to do as agent blazers. You are going to use this agentic technology to do work for you to make experiences like that. And that is what Salesforce Headless 360 delivers for you. It is an amazingly powerful tool. I can't wait for you to learn more about it the rest of the conference. But we got a couple more things to do. So some of you may be thinking, you know, well, we saw, we saw, we saw, you know, like a gold hoodie. But maybe that wasn't enough gold hoodies to be handed out today. And so I said I wanted my share. If I'm going to host this thing, I want to hand one out. I mean, that's part of the gig, right? So I'm going to come over here, and I'm going to talk to my new friend Sarah, who works for ngINJ. And yeah. Earlier, I don't think everybody knew what Engine was doing. So now I want to ask you as the first question, how does it feel to be a celebrity?

You know, I have my bodyguards right here. They like to call them, you know.

Nice.

It's awesome being here. Thank you so much.

That's great. And, you know, one of the things that we've talked about is that a lot of times I see enterprises getting stuck on where to get started, right? It's like, oh, man. I mean, I can imagine if you're a trailblazer here, you've been working in Salesforce for a bunch of years. You see all this, you're like, wow, it's overwhelming. How did you guys get started at Enjin?

Yeah, absolutely. It's always a team effort. So a huge shout out to the broader Enjin team. Without them leaning in and truly trusting the process and the product, none of this would be possible. So huge shout out there. But it absolutely can feel daunting, whether you're releasing an agent that's customer-facing, employee-facing. It can be scary. We've been there and we've done that. So what we did was we took this agent, and we were able to actually understand the customer's issues by looking at our data. And we found a simple use case that we could take and give to the AI, right? A repetitive, a manual task for our human support agents, right? And we would govern, we would use quality assurance to make sure that that agent was actually doing what we wanted it to. And once we developed that trust and that confidence, we were, we were able to scale a lot easier once that foundation had been laid.

That's great. Well, now you get started. I mean, you got literally the poster child of TDX at this point in time. So how did you scale it? Because you had that initial experience. How did you, how did you really get it going everywhere?

Yeah, I think the moment that it went from one single agent to, you know, an agentic enterprise, when we were able to actually take the patterns and the infrastructure that we built for that first agent and apply it to our net new agents. So as we were spinning up agents across the business, it wasn't ever like we were starting from scratch, right? We didn't have to reinvent the wheel. We had these building blocks in place. And so it wasn't just, oh, we built this awesome service agent. It was more like, oh, we built a foundation. And that single shift going from one agent to that scalable ecosystem of agents, it really helped redefine the business and what's possible with AI.

Excellent. Well, I want you to hang right here for the hoodie because—

Wait, I'm actually getting a hoodie?

Yeah, you're totally going to get a hoodie.

What?

Yeah, yeah, absolutely. I mean, yeah, I wouldn't come down and talk to you if you weren't going to give me a hoodie. I wouldn't deek you like that. But, you know, because I threw a fit because Patrick got to give out a hoodie earlier and stuff like that. You wanted to. Yeah, I wanted to give out a hoodie. But to pacify me, they actually let me get out two hoodies. So I'm going to welcome our new friend, Sally. Sally, you're, you're with United Roads. Yeah, give it up. You're with United Road Services, and you're—

you—

I love your story because not only for your company United are you working on, you know, agents for yourselves, you are a huge evangelist, and you're trying to bring the whole community, this whole room with you. It's a lot of pressure on the shoulders, but you're doing it, and you're doing it with grace. Tell us a little bit about the work that you've been doing trying to both evangelize internally and externally the agentic revolution.

Yeah, thank you so much, Joan, for having me here. So I feel like I've been in the ecosystem for 16 years, and there was always changes in like technology and how to achieve my day-to-day job. But recently we are in the biggest change, and with the agent and how to build, and it's a big change how we can use like agents to build and achieve changes. And with that, like, I've been always curious to learn and try to apply it in United Roads and also to help the community learn what I'm learning. And I believe strongly, like, we have to keep learning, knowing our basic skills and how to do our changes ourselves as well, to be able to double-check and have the human in the loop to review the agents and what we are building. And that's why I encourage everyone to keep learning and staying curious and learn about the AI. And today we have a lot to learn. I'm curious myself to learn more about what we just saw in the presentation. And I try to share this with the community as well through Code with Sally. Code with Sally, I started because I love learning and teaching and also I love to share what I learn with the community. And also Salesforce changed a lot a lot of my life and like my family life as well. And we were able to move across world, uh, like with the Salesforce opportunities. So I feel like I, I should give back to the community and help like learn and give back, uh, with sharing the knowledge. Uh, and that's why I started Code with Sally as well, just to increase the impact, like not just through my day-to-day job, but to help others learn as well. And it started as a blog and then I moved to have weekly live sessions, both in English and Arabic. So at least I have two sessions weekly, and the Arabic is to break like the language barrier for other people in other countries to learn in their comfortable language, and also to make it more achievable and like to make like people accessible to learn. I upload the sessions on YouTube so they can achieve like offline learning, and also I share share the code on GitHub so they can check whatever artifacts we did during the sessions. And that's why I feel it's a great experience because I feel the community and Salesforce Ohana is always about the community and how we are helping each other. And I feel I'm happy to do this through Code with Sally. And I'm honored to be part of the community and Salesforce family as well.

Yeah.

It's so amazing. I mean, absolutely amazing. You are an inspiration to us all. I mean, it's seriously super inspiring story. It's really incredible. We're so thrilled to have you part of the community. Where are my hoodies? Oh, here we go. All right. I put the mic down for a second. Yeah, yeah, of course you got to put it on. All right. All right. Thank you both. I'm going to hand that back to you. Thank you so much. Good to talk. All right. I don't know how to follow up. That was like probably one of the most inspiring things I've heard in like 2 years. So like, I'm going to try to get some composure and bring it back. But just amazing job with both of them. Just the work they're doing is incredible. And I think I think you could see, you could feel the transformation. It's like really, really electric. But back to like, this doesn't feel like a Salesforce thing. So I'm gonna bring it a little bit grounded back to you on like, you probably, I don't know, maybe work in Agentforce Sales. So we are gonna have some Agentforce Sales for you. So visit the campground and check that out. I don't know, maybe you work in Agentforce Service. We're gonna have something for you too. Here's a QR code. Go check out Agentforce Service. Some of you may be using Qualified, like our new acquisition. Qualified in the house. Go check that out as well. Agentforce Supply Chain. Super cool Agentic implementation here. Amazing, amazing stuff. Hit the QR code. Go check that out. Near and dear to my heart, Slackbot. Incredible stuff. You saw it rear its head many places in these demos. That's because we live on it at Salesforce, and it's absolutely incredible. So go beat your personal work agent. The grounding, Data 360. Again, it was here. It's in there. Go check out more stuff. There's the QR code. And then think about the insights that you're going to get with Tableau. So incredible amount of stuff. All your favorite Salesforce greatest hits are still there. But Agentforce Headless 360 is a peek of the future. And the future is now. So please take a look at the screen. Look at all the rest of the stuff we have coming up. You're going to get some flow, right? I mean, what else do you want? This is Agent Force Headless 360. Thank you for your time. Please enjoy and please check it out because it will absolutely change the way that you do work. Thanks.