Build and Govern the Agentic Enterprise | IT Keynote, Dreamforce 2025

Please welcome senior vice president, product management.

AI app development, salesforce, kushwant singh.

All right.

Good morning, friends. It is 9am the day after our big party, and so we are thrilled to see so so many of you here ready to join us kick off day three of Dreamforce. So let's get going. Now, I've been at Salesforce for over 10 years and first and foremost, it's great to see so many familiar faces in the audience. Now together, we have effectively navigated through many prior technology revolutions, whether it was mobile, social or predictive AI. And today, we stand in front of another revolution, that being agentic AI. Now, all week you've heard about the agentic enterprise and how transformative it can be for enterprise productivity. I've spoken with a number of you and heard your questions. Questions like, what does it mean for an agent to proactively engage? How do I build it, test it, configure it? How do I ensure it has the right guardrails? What do I need to do differently so that I can help lead my organization to become an agentic enterprise? Now, with so much changing around us so quickly, it's natural to have all of these questions. But like any prior technology revolution in the past, we here at Salesforce, we're committed to helping you cross this divide as you work towards becoming an agentic enterprise. Now, on the subject of crossing divides, we here at Salesforce are also going through our own transformation. Our help site that you see over here today, it has served over 1.8 million AgentForce conversations in seven different languages. It's helped us save over 100 million in cost. Similarly, within Salesforce for all 70,000 employees, we are revamping the employee experience around Agentforce and Slack for a number of jobs to be done, be it filing an IT service ticket, filing my expense reports, the whole thing. In fact, on my teams, we are using AgentForce to help migrate all of our aura components over to lightning web components using AgentForce versus trying to do it manually. One by one. AgentForce has truly extended and elevated every team, every employee at Salesforce. And we want to be sure that you leave today knowing the steps and the capabilities you have at your disposal to build and govern the agentic enterprise. Now, over the years, we've built a lot together and we truly appreciate your trust, your partnership. You've trusted us with your workflows, you've trusted us with your custom applications, you've trusted us with your metadata, your data. And the beauty of all of that really, is that it's already Setting you up on your way to become an agenti enterprise. Your flows, your apex. These are prime candidates to become actions for your agents. Your custom apps. These are the appropriate surface areas for you to introspect and look at which use cases could benefit from a conversational agent, a background agent. The data, the metadata that you have already invested provides your agentic apps, provides your agents with a unified context upon which they can operate. On the subject of agents, close to 200,000 agents have already been built and we have a ton of developers who are already benefiting from AgentForce vibes to drive their productivity. And so you're already on your way. And to ensure that you accelerate, we at Salesforce have to innovate. And this week we introduced AgentForce360. AgentForce360 comprises of three key aspects. The first one is what we call our customer 360. Think of customer 360 as a set of out of the box agentic applications and solutions for a variety of different verticals and industries which you can use to get started quickly, flexible, customized to your needs. The second part is the data360 and data360 is what gives you a unified view of your data and and your metadata, whether that is within Salesforce or outside of Salesforce. And then finally we have the data layer, or rather the agentic layer. And this is what provides you with the building blocks to build, test, deploy, monitor, observe, improve all of your agents, your agentic applications. And to see all of this in life, let's take a look at the journey that Pearson has been on.

You're born to learn, you're always learning.

Throughout your whole life and Pearson is directly involved in supporting all of those different aspects of learning. Pearson's a 181-year-old company helping millions of learners achieve the life they imagine. From virtual schools and higher education to our fastest growing businesses in enterprise, helping retrain and reskill within the workforce. But we're facing such dramatic innovation and change in the skill of what jobs are needed. Before in our partnership with Salesforce, we didn't have that deep contextual understanding of our customers. And now we're building that. We knew that we had work to do to be able to take advantage of agentic AI. Building our petabytes of data and history to up our game on data foundations. All of that is underpinned by Data Cloud and Agentforce. Developing a customer360 and really personalizing the learning experiences themselves is a big step forward for us.

If we don't have proper data Data.

That we can trust. We don't have anything. Data cloud is truly the foundation.

We are multi cloud and we're devoutly hybrid AI. The fact that Salesforce can sit at the heart of that and our Salesforce agents can work with agents from other platforms, that's gonna make it so that we can move faster and drive innovation in a much more impactful way.

There was no reason we wouldn't use Salesforce's agentic platform to be part of that journey. With agent Force, we're taking that next leap.

There is huge emotional weight to a high stakes learner who has a case open with us. It's unacceptable to put a customer ON.

Hold for 10 minutes when you could.

Have had an agent give them the answer instantly. We handle nearly 400,000 customer inquiries every year. And within the first six weeks of implementing the Salesforce agent, we've seen close to 24,000 call deflections. Changing the experience not just for the customers, but for our own employees and serving those customers and realizing that the investments are going to pay off massively. That's the power of a human plus a digital worker. I think it's a little bit of a miracle that we have been able.

To give one Pearson face to the customer.

There's no way we could have done that without Salesforce. I have been brought to tears to hear stories of these learners, the career changes that it unlocked and then the pride that we have to be able to support them in doing it. It's hard not to be emotional about it. This is not just innovation for innovation's sake. This is understanding that agentic AI is fundamentally chang the nature of the digital world. And we want to be at the heart of it, just like Salesforce is at the heart of it. And we're going to drive forward together.

All right.

An incredible journey. And to learn more, please welcome David Treat, Chief Technology officer at Pearson. How about a round of applause? Welcome to Dreamforce. Dave, how's dreamfast treating you?

It's been amazing. I think my brain is stretched. My heart is full. My ears are ringing. From Metallica.

Excellent. Excellent. You were on that video.

Sorry?

You were on that video with the Metallica video. No. So, Dave, look. Incredible journey. What are you most proud of and what were some of the challenges?

I'm really most proud of the team and some of them are here. Super proud to see them in the video.

It is.

And it's not just our team. I ran into Claude Sutterland outside Starbucks this morning. Your forward deployed engineer that's helping us it's the pride I have in watching our joint teams work together to do something that's really hard. We have a data infrastructure and an application infrastructure that is modernizing and being able to plug agents into data sources and applications. It's hard. It takes hard work, and we've done it successfully. That 400,000 number was off. We have millions of customer contacts. I was referencing just one specific in case anyone's questioning, but it's been a really powerful experience in working together and partnering to do something that's difficult, that has such a dramatic impact for our operations and our customers.

Now, I appreciate you keeping it real and saying it hard. I mean, even on our side, what we showed with the Salesforce help site took weeks and months of iteration. So I appreciate you keeping it real. So, Dave, as you think about it, why Salesforce as a partner, we want.

To work with some. We're working with a set of partners who are as serious about this as we are. We are not taking half swings. We recognize the dramatic power and impact of what agentic AI will have to our business, to our customers, experiences, to our customers, businesses. So we just appreciate that you're taking that full swing with us and the depth of our partnership and history together. You know us, we know how to work together.

Right?

There's magic in that.

No, we appreciate the partnership. Now, as you think about this journey that you're on, where does security stand for you?

We have the privilege and the responsibility of serving learners of all ages. And so if you think about The K through 12 student who needs to have a clinical assessment and the sensitivity of the data around that, the parent child relationship, you go into higher education, you go into enterprise. We're serving learners in so many different experiences. The trust that they place in us, the trust that governments place in us, school districts, department heads, parents, the individual child, of course, we take that. It's at the heart of everything we do. Our focus on security is pervasive through the whole experience. My ciso, it is the utmost importance to us. And again, we appreciate how you also feel the same way. And lean in. And we're talking about the security infrastructure right from the start and the design of what is an increasingly complicated, you know, architecture.

Plus one to that. I mean, I've spoken with many IT leaders. It all starts and ends with security. Now, as you think about your, you know, weeks and months ahead of you, what are some of your top priorities?

Like I said, we're taking a full swing. So our priorities break down into three categories. We are Driving AI, agentic AI into the products we're giving out to customers. And we're about 18 months into that and getting incredible data points around the effectiveness of changing the nature of learning experiences to be not teleporting to the answer with AI, but actually using AI based experiences to scaffold an experience and actually engage the brain. And the data points are amazing where we're getting four times the engagement. 85% of the students that are using our tools are getting the grade they want or better in a higher ed context. So we're getting incredible feedback. So customer facing product based on agentic AI experiences. Priority, priority two is we're going to fundamentally change the nature of the software development life cycle and so adding agentic AI to automating that and across all of its dimensions. So AI for engineering, priority two and priority three is agentic operations, looking at all of our internal workflows across all of our corporate functions and our business units. So I have three separate teams, each with that dedicated focus. You're in the mix with all of it and we appreciate it, but it really is the priority is to take that full circle swing and not just go for the safe, easy internal use case, but be out with customers, be really fundamentally changing the engineering and how we operate.

Excellent. Well, thank you Dave, for all that you and your team are doing. How about a round of applause? Thank you for Dave and his entire team. Thank you. Now, if you think about it, interestingly enough, both Pearson and Salesforce are great examples in how agentic enterprises still require the same fundamentals that you value in your IT ecosystem today. You want a unified yet open platform, not just for your agents, but for your apps, for both of them, whether they were built at Salesforce or not. You want all of them. You want these apps, you want these agents to be able to work together. You want them to be able to work together on a unified context of data and metadata so that they can effectively accomplish your business workflows. Similarly, you want a robust DevOps lifecycle. And as it pertains to security, governance, compliance, trust, this has always been top of mind for you. And in an agentic enterprise, this goes on overdrive. So what we're going to do now is we're going to double click into each one of these fundamentals. We are going to show you the capabilities that you have at your disposal within each one of those fundamentals to build and govern the agentic enterprise with confidence. And to kick us off, please welcome my partner in crime and distinguished architect Leo Tran.

Thanks Kush. So I've been an engineering architect at the Salesforce platform for almost 17 years. I know how important it is to all of you that Salesforce is an open and unified platform, a platform that works across your systems and the entire enterprise architecture. Also heard from many of you that building for your enterprise today can feel like solving the ultimate jigsaw puzzle. The many pieces, those are your many systems, maybe in a few data lakes, all built over years and added by multiple teams. So how are you supposed to tackle this? Well, I know when I start solving a jigsaw puzzle, I refer to to the picture on the box. And that's what Salesforce can offer you. It can be this opinionated guide. It can help you connect pieces and can even offer the ones that you might be missing. And so, to start, let's get back to the fundamentals. The first challenge you've always needed to solve is to build that accurate context, that holistic understanding of everything that's happening in your enterprise, from how your sales teams are working with their customers to how your customers are engaging with your product. And this is where Data 360 can help you build that accurate context by unifying data. It can also connect your existing systems and existing data lakes with zero copy federation. And so you don't have to pay that cost of moving all that data around. Now you have that accurate context, you understand what's happening in the enterprise. How do you take action on that context? Well, if you have a handful of systems, you can use capabilities like external client apps, external services, Salesforce connect to integrate Salesforce and your systems. When you go from a handful of systems to many, maybe hundreds of APIs, we all know that quickly becomes a tangle. Frustrating for humans to understand, impossible for an agent to navigate. And that's where the power of Mulesoft really comes in. It can help you untangle this knot of APIs into a meaningful set of actions that are ready for workflows and agents to execute securely and reliably. So that's the fundamental strategy. Meet accurate context, meaningful actions. And to see this come to life right now, please give a big welcome to Senior Director of Product Marketing, Kat Su.

Thanks, Leo.

All right, so let's start this demo first by introducing, introducing you to our demo company, Connected Executive Education, or CEE for short. And just like Pearson, CEE is an education company that delivers personalized experiences to every single one of their customers. You can see what this looks like on the screen. It welcomes Sarah Johnson, and it also has this incredible learning journey agent that's giving her personalized responses based on all the data Cee knows about her. But how did Cee create this? Well, we all know that personalized experiences start with your data. And just like so many other companies, Cee had their data scattered everywhere. Learning records in their LMS systems, billing information in their finance tools, classroom history stored in data lakes. All this information is really great. But true power comes from being able to connect all that data and start to deliver a more robust view of their customers. In order to do that, though, they needed a connected foundation. And that's where Data 360 comes in. Whether it's their certification data in Amazon S3 or classroom training history in Snowflake, Data 360 allows them to connect all of that data in one place and start to map all of that data together so you can start to build that unified view of your customer. What's really incredible, too, is that, as I mentioned earlier, Cee had an existing investment with Snowflake. They have years and years of classroom training history stored in Snowflake already. But thanks to the power of Zero Copy Data Federation, they're able to access all of that data right where it lives today. No copying, no duplication of that data. On top of that, they can access all that data back in Data Cloud and connect it to all those other data sources. So now they have this really powerful view of all of this data on their learners without having to move any sensitive data.

All right, let's take a moment. Cat, what you just did by connecting structured data, that's like connecting the edge pieces of the puzzle. Huge milestone. You have the frame, and now let's start filling in that trickier middle. It's a little more complicated, but it's where the picture really starts coming to life. And for Cee, that means processing their unstructured data. All that critical information locked away in digital documents and scanned forms. I'm sure many of you can relate. All of you have critical information that isn't just queryable as records in a database. So, Cat, how's CE going to connect the next set of pieces?

Yeah, let me show you, Leo. So here you can see an example of some of that unstructured data that CE has. This is a classroom enrollment form. And as you can see, it's just a static PDF. This is largely unusable. But you can also see that it has so much rich information that's so valuable for a company like Cee. You see the job title, course name, course id. But now, thanks to Document AI Cee can upload all of this unstructured data in bulk and document AI processes all that data and uploads it to directly into Data 360 as live records. So now not only does Cee have all that structured data that we connected earlier, but they also have all of this rich, unstructured data to create an even more powerful view of their customers.

Yes, that's what I'm talking about. And so what Kat just showed, that's also what Salesforce did. So remember when Kush brought up the help agent on help.salesforce.com well, Salesforce also used Data 360 to process almost 200,000 pieces of help content. Think release notes, knowledge articles, the help documentation itself. Having this accurate context of structured data and unstructured content was critical to really uploading a help agent to answer the questions that our customers were asking.

Yeah, definitely, Leo. So we all know that data is kind of just step one in the journey here, right? You have all this rich context, but what good is that if you can't actually take action on that context and extend that action beyond Salesforce and across your entire ecosystem? Well, you're all in it. You know that doing that work takes a lot of time. It's manual. And in the age of agentic AI, you don't have time to waste. You need to move fast to stay ahead of your competition. And that's why mulesoft is so powerful. Mulesoft has a vast library of out of the box connectors that allows you to get started fast. So Cee has all their billing information today in NetSuite, but you can see here, with just a few clicks, they're able to connect all of that data right into a flow. This is incredible, but it just starts there. Netsuite is just one example. CEE has dozens of APIs that they want to connect. And you heard Leo talk earlier about how so many customers just like you today have this tangle of APIs that are so difficult to manage. But that's where Mulesoft is even better. Because not only does it allow you to connect all your APIs, it also helps you govern all of them. It provides this control plane that allows you not just to connect, but to see all of your APIs and ensure they're secure and standardize access across all of it.

That's amazing. What Kat just showed is the fundamental, the foundation of an accurate context. And now we have a set of meaningful actions. Amazing. Please, everyone give a big round of applause to Kat. All righty, so we have Meaningful actions. What's next? How do we empower agents to take action with them? And so the challenge is, how does an agent know what's available? For years, developers like you have created open API specifications to communicate to other developers what APIs are available and how to use them. Well, there's an emerging standard called the Model Context Protocol, or mcp, that is how developers communicate to agents. MCP enables an agent to discover what APIs are available and which APIs should be used. And MCP is natively integrated into MuleSoft. So that meaningful set of actions you have in MuleSoft, you can add this additional layer of MCP metadata right onto the actions that agents should know about. And once you're able to build one agent that can take actions with mcp, you're going to want a team of agents being able to collaborate and focus on specific industries and domains like inventory management or case resolution. But for a team of agents to truly collaborate, they need to speak the same language. And this is where agent to agent or the A2 protocol becomes really important. A2A is how an agent can collaborate each other, declare the capabilities they support, delegate tasks and share information. And 8 will also be natively integrated into MuleSoft, which is where you can see MuleSoft can be your control plane for your entire agentic enterprise. It can help you untangle your APIs, manage this growing set of MCP based actions and your growing team of agents. And so let's recap the fundamentals of what we just showed here. With CEE, they now have an accurate context built with Data360 that unifies all of their structured and unstructured data. They've leveraged mulesoft to untangle that knot of APIs they have into a meaningful set of actions. And now they're ready to build agents with MCP and A2A. So with the Salesforce platform, what you're seeing is how Salesforce can help you connect your existing pieces, offer you the ones you might be missing, but most importantly, give you that picture on that box, that opinionated guide for how you can build agents faster and better than any DIY approach. So now if we go back to that picture in the box, we can figure out which section to focus on next. What happens when you have hundreds of apps now, hundreds of APIs, and you're going to have a team of growing agents? Well, to talk about how to maintain a well governed app and agent lifecycle, please welcome the COO of mulesoft, Ayeong Ahn.

Thank you, Leo it's so helpful to see how you can unify application agent and data together as the first step to becoming an agentic enterprise. But what's next? I'm going to give you a hint. It's something that's top of mind for both you and your CIOs. It might also be on the slide. Governance. I talk to IT organizations across every industry and governance is always top of mind and always a massive challenge. Then you throw agents into the mix. Now the issue is on a whole another level. Because the lack of governance isn't just about big issues like outages and breaches. It's also about not knowing what answers your agents are providing. It's about not knowing that there's an issue until your customer tells you. Ultimately, the lack of governance means losing your customers trust, which could lead to losing them altogether. But don't worry, that's exactly why we're here to help. In true Salesforce style, we're going to make this super easy for you by building in governance at every step of the process so you don't have to DIY and figure it out yourself. So to do that, we're going to go back to Cee. Cee wants to build a personalized agent first experience for their learners. And as an agentic enterprise, they know that they can use AgentForce for that. So to show us how this is going to work, I'd love to introduce John Bello, Senior Director of Product Management. Hey, John.

Hey.

Hey, how are you doing? Thank you all. Great to be here. All right, we are going to do the right thing and we are going to start in a sandbox because sandboxes are those safe environments outside of production and that's really where we should be building our agent. But note one important thing here. There are actual learners in that sandbox already. And this is a developer sandbox. It doesn't come with any data. So we seeded the sandbox with synthetic data. Those are completely fake students, but they match the data shape in production. So we will use this data later to actually test our agent once we start building it. And to start to build it, we'll go to AgentForce Builder. And this is the same AgentForce Builder you already know. You're already using it today in completely local ways to build your agents. But now we can do so much faster because I just need to describe the agent I want and Agent Force Builder will instantly generate topics, actions, instructions that I can use to set up my agent and the whole process gets so much faster. I just need to add Additional capabilities. And when I'm happy, I can visualize the agent here in my conversation preview panel and ask questions and pretend I'm a learner, see how that agent would interact with me.

Awesome. Now that I love that it's fast, easy and low code. So now that the agent is built, we got to test it. Remember earlier I mentioned the lack of governance is not knowing what answers your agents are providing. I'm going to help you with that. Testing your agents is imperative because agents are non deterministic. That means you don't really know how it's going to handle unexpected questions unless you put it through some rigorous testing. And that's exactly why we have the Agent Force Testing center here. You can run your agents against real life test cases, see how they respond, and if you don't like their responses, you can augment the instructions to train it some more. So John, can you show us how this works?

Let's do it. All right, so we have a number of agents Force Testing center test cases that we are running against our agent. And now we're going to take a look at our results and we can see that they are providing responses that are complete, coherent. We are happy we can go ahead and start the process of actually releasing our agent. This is when I pause to say that my teams have been hard at work at the tools that I'm about to show you. So thank you to my teams in India, Argentina, in the US working hard to help you, our customers, deliver innovation faster to your users. With that said, let's go to DX Inspector right here. Feel free to do that. Keep clapping. Keep clapping. DX Inspector. This is where I go to identify what has actually changed in my sandbox. New thing is that we are now able to select and commit my changes, my learning agent, all of its configurations, instructions, directly into our release management tool. And this is a tool your teams already trust reimagined to meet the needs of an agentic enterprise. This is the next generation of DevOps Center. And DevOps center is now deeply integrated into our platform, which means it's lightning fast, easy to set up, easy to use, and completely free and available to all Salesforce customers. No exceptions there, right? And I can use this view, this pipeline view here to track all of my changes as the agent flows from left, my development environments all the way to the right, my production environments at every single step of the way, at every stage. I can run automated tests with quality gates and ensure that my agent has been thoroughly tested with a range of free tools. Maybe Even partner commercial tools. And I only release it when I'm really, truly happy. So DevOps center really gives you this completely local way to have governance at every single step of the way as you manage your releases.

John. But I noticed that you didn't mention something that's really important that's going to change the application, DevOps and agent lifecycle. You guys want to know?

Let's do it. Diane.

You guys want to know? I am so excited to announce the brand new Agent Force Vibes. How many of you love Vibe coding? Yeah. And how many of you are googling what that is? No one's going to tell me. Well, AgentForce vibes completely transforms the application, Agent and DevOps lifecycle by making it agentic. Now, using natural language prompts, you can ideate, build, test and deploy and manage your end to end lifecycle, making it a dream come true for not only your developers, but also your release managers, your QA and more. My personal favorite thing about AgentForce vibes is that it already understands the Salesforce platform because it comes with the MCP tools built in. For example, it knows DevOps center for release management, it knows code analyzer for testing. And of course governance is built in at every step of the way with the Salesforce platform and the trust layer. But that's not all. We're bringing the agentforce Vibes experience directly into your flow of work. That's right. Now you can manage the full end to end DevOps lifecycle directly in Slack. And as you can see as it's actually moving through DevOps pipeline, you're seeing the notification and alerts directly in the Slack. It's everything you and your team have wanted to work better and faster.

It's practically perfect and I love it.

Ayang, I'm literally mind blown, right? Agent Force 5 in Slack, able to communicate with DevOps center and operate it fully end to end. This is awesome. Practically perfect. But nothing in it is ever perfect. You guys are always pushing boundaries all the time, delivering innovation to your users. And when we move fast, sometimes we might risk breaking things. So we need a backup strategy. That's why Cee set up backup and recover so they can restore any data fast, with precision, ensuring that their agent always has that reliable data foundation that they need to support their clients. That's only part of it though, because we also need to ensure that the agent has access to fresh data. So with archive, old records can be automatically stored away so that our agent is only ever pulling from up to date data. And here's an example of AgentForce that actually has access to a lot of cases, including cases we no longer handle. Now we can set up archive policies for data and ensure that our agent is only using up to date relevant cases when supporting customers. But importantly, you see those archive cases over there? We always have access to this information, to this data in Salesforce for compliance purposes.

All right, John, let's take a quick pause and do a recap. So we built our agent by doing the right thing. We started in sandbox and then we tested it to make sure that it's giving the right responses. And then we pushed it through the DevOps pipeline and along the way we got alerts in Slack. And last but not least, we just saw that our agent always has reliable information through backup and recovery. So now that our agent is up and running, how do we continue to ensure governance?

Well, we have the new agent force observability, right? So we can go there. It will give you a number of key insights, roi, adoption rate and so on so we can understand how our agents are actually performing. And I can even drill down to specific sessions, specific user sessions and see which kind of topics, which kind of interactions is actually using for particular sessions. And using this data, I can see what's actually working well and where we need to adjust. Now we all want agents that are working well, of course, I mean that is a given, right? But we also want to keep our costs under control as we add more agents. And this, I know you care about this Ayang, right? And for that we have Digital Wallet. It helps us track consumption across our usage based products such as data360 such as agentforce. I can see how agents are actually using resources, forecast, demand. It just gives me a complete picture on how my resources are being spent. And importantly, I can also set up alerts so that my admins don't even need to come to this dashboard. They will get notifications that consumption has reached certain limits so they can take proactive action and stay atop of their budget.

John, you are speaking my love language. There's nothing COOs love more than an alert that helps my team stay on budget. Amazing. So now that our agent is up and running, how do we make sure that we fulfill the experience we really wanted for our learners? Well, to do that, our agents need to be able to work together, connect with each other to answer the questions and challenges that our learners have. And of course we need to do this with governance and visibility built in, which is exactly why we built and I'm so proud to announce Mulesoft Agent Fabric. Mulesoft Agent Fabric empowers you to manage, govern and observe all all your agents across your ecosystem, no matter where it was built. You can also add on additional layers of security or policy using policies or guardrails that were built specifically for agent to agent or agent to system interactions. The magic of it all really comes to life as these things start to connect. Because you see your agents, your LLMs, your MCP servers, how they're connecting, how they're interacting and how they're performing. With agent fabric, CEE can rest assured that they have the proper level of governance and visibility into all of their internal and external agent interactions to do the right thing. John, can you show us how this works?

Let's go to Agent Visualizer. Then we have this awesome map that shows me all of my agents see there learning agent working with the upscale agent working with not notification agent collaborating together to deliver complex cross system tasks. Thanks to MCP and thanks to A2A. And because I have this visibility, I can now intelligently orchestrate my agents routing tasks to the right agent at the right time. But without governance, agent orchestration turns to chaos. So that's why we have agent fabric delivering the guardrails that enterprises need and ensuring that across every agent interaction, there are security policies, compliance and controls applied to all of it, regardless of where that agent might live.

Thank you so much, John, everybody. Give John a round of applause. So you just saw how to ideate, build, test, deploy and manage all of your applications and agents the salesforce way. That means governance is not bolted on or an afterthought. It's with you every step of the way. But as an agentic enterprise, you also need trust. And that trust comes from knowing that all of your agents and applications are operating with the highest level of security. So to teach us how to maintain enterprise grade security, I would love to welcome to the stage SVP of Product Marketing, Josie Trials.

Thanks, Ayam. When Dave said that security is his first thought, his last thought, and every thought in between, I saw a lot of nods in this room. And that's because I know you all take data security very seriously. Because when we're talking about data security, what we're really talking about are our customers, our colleagues, our friends, our family. This is personal. So let's dive into the last piece of the puzzle and see how together we can deliver built in trust, security and compliance. The shift to the agentic era brings two things into sharp focus. The unprecedented opportunity and the unprecedented risk of autonomous AI. The opportunity is Clear AI allows us to unleash unfathomable productivity faster than ever before. But the risk is also just as real. And in fact, the cost is incredibly high. The average cost of a data breach is now nearly $5 million. And that's a number that directly impacts your budgets. But more importantly, it impacts the trust that you're responsible for maintaining. Securing the AI enterprise really starts with a partnership. And at Salesforce, our security is built on a shared responsibility model. One that clearly defines the roles we both play in protecting your data. We provide that trusted secure platform and that includes things like our trusted scalable Hyperforce infrastructure and agentforce trust layer, which governs interactions between model providers and scans for things like toxicity and bias. And then you do things like take our very strong built in controls and use those with your unique business context. And this is things like turning on IP allow lists or setting up mfa. Okay, this is really important. If you haven't set up MFA or IP allow list, this should absolutely be the very first thing you do when you get back on Monday. Together, that creates that secure foundation for everyone. But for those of you who have complex regulatory environments or who are looking to automate some of this manual process, Salesforce provides sophisticated add on tools to help you scale your security. And this is things like SHIELD for event monitoring or comprehensive encryption. And this year we introduced OWN into the family and this increased our suite of tools, but it brought us also best in class tools like backup and recover for comprehensive data protection. We just recently announced a new partnership with CrowdStrike and this is going to bring CrowdStrike's leading Falcon platform into Security Center. So this means you're going to have a single place to see all of your threats and your permissions right in Security Center. And this is across your entire tech ecosystem, not just those things that are living within Salesforce. So to show us all of this in action, I'd like to introduce my colleague, Susie Compton, VP of product for trusted services. Suzy, let's see a demo.

Yeah.

Thanks, Josie. Hi everyone. I lead half the trusted services product teams and I am so excited to show you the incredible innovation we've been working on. And from my conversations with you all over the past 13 years, I am very confident you're going to be just as jazzed as I am. We saw earlier how Cee built agents with governance. But the next step, the next critical challenge is trust. And they are asking the same question all of you have been asking, which is how do they make sure that sensitive information does not leave Salesforce that's where the trust layer that Josie mentioned earlier comes into play. But what is it?

Well, it's how you can protect your.

Data in the context of AI to make sure that that information is never leaving Salesforce boundary. And as we can see here, you have the ability to configure on an individual level what fields you'd like masked personal information like email address, phone number and even encrypted fields so that it's masked before it goes through the LLM and into your AI prompts. We've made it easier so you can mask based off of the compliance categories that your company has created. We know that CEE has to meet GDPR requirements, so we're going to go ahead and toggle that on. But they, just like all of you, also have data flowing through other external systems using data360 and they need to protect that too. And that's where the magic of AI data classification comes in. Built right into your Data360 where you've already been doing the manual work of data tagging, field tagging. I think this is a use case of AI that we can all get behind. Because what if we could just press this magic button that has stars on it and voila, the power of AI has tagged and scanned automatically all of that external certification data that has flowed through. And all that's left for the admin is to select the field, view the suggestions and approve it. This is the AI classification and tagging that you guys have been asking us to do for a while now. But let's take a step back because security is more than just the fields, it's also about the broader view. It's about the entire environment and how you're configuring it and how you're monitoring it. But you already know that, and you're already likely familiar with this view of Security center, where you have the ability to manage your security across multiple orgs all in one place, so nothing slips through the cracks. And you can manage privileged users for privileged user access for over 80 perms, including who has access to data, Cloud Data360 and how you can monitor and track your agents. Plus, the Security center that you know and love is now enhanced and includes Security Insights, which was part of our own acquisition earlier this year. You can think of Security Insights as you're a security expert, which provides detailed insights into key risk lenses that is going to give you the ability to analyze, proactively, monitor, prioritize and even remediate threats. So let's click into this Data loss prevention lens because your infosec teams now more than ever are probably focused on Salesforce and this is going to help guide you. We can see that Cee has done a fantastic job of tracking their high risk fields, but we can also see if we scroll down a little bit that there is a high number of users that can modify or export all of their data. Now that they know this, they can go back through Security center and make some of those changes. So they've got this fantastic framework and foundation for security. But wouldn't it be amazing to have an always on agent that is logging and monitoring your incidents and issues before they become even bigger?

Suzy, that would be incredible. And that's why I am so excited to announce the new Security center now with AgentForce and I'd love to give you just a little behind the scenes on this one. The Friday after Dreamforce last year, a small group of us met in the Salesforce tower and we were talking about agentforce and how cool it was and we were thinking how can we bring the power of Agent force to it. And that's where this product came from. Because now with AgentForce Insecurity you have a 24.7ai security agent that is enhancing your security operations center and helping you take that manual threat hunting and turn it into something that is guided and automated. Suzy, I have been waiting for this moment. Will you show it to us?

Yeah. I love that. Very true story. So we all know the manual effort of investigating and doing reactive investigations and how long that takes. But what if we could flip that and blow your security team's minds? Starting with Slack, a proactive detection has been done by the agent in the background and it's notified your team in Slack so they have that immediate action alert. And as you can see, this message in Slack is showing you what the issue is when it happened. A succinct summary and the ability to learn more, which we're going to click into and it's going to take us to this dashboard within Security center which is giving an overview of all of the active investigations in their status. But Cee wants to know exactly what happened with the session hijacking. So we're going to click into that. But remember how we said nobody's got time to dig through logs, nobody's got time to figure out configurations, nobody's got time to do all of that. So we're actually going to go over to the right hand side where you can see the agent. We're going to say hey, who and what are the objects that Dan has been able to access and boom. The agent on the left hand side has provided us a list of the objects just like that without you diving in. Incredible.

Right when seconds matter, getting these alerts right in the flow of work that is impacted. But we also know that many of you spend a lot of time thinking about complex requirements from GDPR to CCPRA. And that's why we have also brought AgentForce into Privacy Center. And you can think of this almost like a Rosetta Stone. It's going to take those complex frameworks that are often changing. It's going to look at your unique. Org context and then it's going to give directed, actionable steps that your admin can take to remediate any of those risks. Suzy, let's see this one.

Yeah, absolutely. As you've probably figured out by now, our teams love delivering innovation into the products that you already know and love. So we saw earlier CEE has GDPR requirements, right? So they're going to go into Privacy center Now enhanced with AgentForce where the agent has been running in the background to surface some potential issues related to gdpr. But it's not just going to find the problems, it's going to help solve it. So we're going to click that resolve button where the agent has provided more information around what the issue is and the solution. So as we see what the solution is, the agent has actually provided a suggestion as we are about to see what a potential policy could be based on the GDPR framework. And one of the suggestions it's saying for a policy is to delete or remove inactive records from production. So all your admin has to do is review and confirm it because this is going to be the way that your security and privacy and compliance experts that know the regulations so well but not Salesforce are able to communicate with your salesforce admins that know Salesforce really well. And as you can now see on the top right hand corner, that compliance score has automatically updated and increased. And as everything that we're doing with agents, when you scroll down you can see that the agent activity has been tracked and logged so that you will always be able to have that at your fingertips. With Security and Privacy center we've completely changed and redefined how we are delivering AI to you.

Awesome.

Thank you Susie.

So now we have that last piece of the puzzle. How together we can deliver that built in trust, security and compliance. It starts with that Agent Force Trust layer and Data Governance360. It extends to Security center with no more manual processes but these automated guided processes and ends with the Agent Force and privacy center. Because admins shouldn't have to be regulations experts to be able to keep your org secure and compliant. And now you have everything you need to confidently guide your company into the agentiq era. Kush, what do you think?

Amazing. Thank you very much. Excellent. As we wrap this up again, I want to draw your attention. These are the three key fundamentals within each one of the fundamentals the capabilities that you have at your disposal to effectively build and govern the agentic enterprise. And obviously there is always one more thing. This week we announced AgentForce IT service. So the same goodness that we use within Salesforce for 70,000 odd employees we want to bring to our customers so that you can help drive faster resolutions, reduced IT costs. Please go ahead to the service cloud booth to learn more. And again look, the day is just getting started and we also have shared a lot with you. We want you to be able to take all of this goodness back to your teams. Please feel free to scan this and to get a guide that you can share with your teams. Lots of phenomenal content for you to spend the day in. We've got two keynotes in particular the Mulesoft keynote in a few hours time and after lunch Security keynote. Everything starts and ends with security. Once again, thank you very much. We really appreciate your partnership and your trust and here's a coffee on us. Your feedback is much appreciated. Have a lovely rest of Dreamforce.