Scale Mission Impact with Data and AI | Agentic Government

Welcome to AgentForce World Tour DC. So there's been a new topic of conversation at my dinner table. I have a son in college, and every time I get together with other parents, the question is always the same. What major is AI-proof? Then I go to work and the question follows me there. Will AI take my job? This question looms large for all of us, but this is what I firmly believe. The skills that make us human are exactly what will win in this new era. Empathy, judgment, the ability to navigate a crisis. These aren't just soft skills. They're the mission. But the fact is that expectations are evolving and budgets and staff remain fixed or even shrinking. And that's the challenge. 81% of constituents say that they want more. Constituent expectations are rising, and over 70% say that they want a personalized experience. But we simply cannot do more. We're already at capacity. Instead, we need to do things differently. We need to offload the work that should be automated, the data entry, the status updates, document generation, and give that time back to the mission. The fact is, government teams are already using predictive, generative, and agentic AI. Think about your city services office. They're using predictive AI to anticipate a spike in cases before it happens. They're using agentic AI— I'm sorry, not yet. We're using generative AI to prefill complex forms and summarize years of case data. And now they're layering in agentic AI. A digital workforce to automate the entire process end to end. Agentek Enterprise for Public Sector is our mission-ready, unified, secure platform designed to turn that AI intelligence into real work. It starts with your system of context, bringing together all of your trusted data. Then you have your system of work. These are your Customer 360 applications, all of the knowledge about your business and how it operates. Then you have your system of agency. This is where you build and deploy agents across every channel and application, and it all comes together in your system of engagement where humans and agents are collaborating in the flow of work, unlocking capacity to deliver mission speed and the services that constituents expect tomorrow. This journey isn't about future-proofing a career against AI. It's about embracing and maximizing your impact with it. Last year, the United States Postal Service joined us to share more about their now and future journey with AI. Take a listen. Oh, that's great to hear. And actually, on AI, you refer to AI as the now and future solution. You touched on a little bit But can you share a little bit more about the approach that the Postal Service is taking?

Yeah, you know, now is coming soon. In the summer, we're going to be switching our call center platform to a cloud-based application where we have natural language interactions instead of a menu-driven IVR. We're going to move it to, hey, in a few simple words, what can we help you with? That'll be the start of our AI journey. But we're looking at agentic AI and agent force and starting what we call kind of a crawl, walk, run approach. We are a government agency. We have privacy concerns. We have image and brand issues. We want to make sure it doesn't run wild. So we're going to start with the crawl and use our 1,600 agents as an incubator. They will then have the natural language. Hey, what is the issue that the customer— go through that tremendously large knowledge base, synthesize responses, And then we'll move to adding additional data sources to be able to strengthen the next best actions and resolutions. And then we'll start to walk and then move that to the broader community in the Postal Service. Another 60,000 Salesforce users that resolve complaints. And then the run is we turn that around and let— through chat functions, we'll let the customers enter in and then work it that way. So we're really excited about it. But we're going to work in that crawl, walk, run, and that's going to start here soon this summer.

That's fantastic.

Crawl, walk, run. A year ago, Mark McCrary grounded us in that approach, and he was right. You know, I was a federal CIO and in government long enough to know that that's just how you adopt new technology, right? Ha, but Agentic AI flips that model, and that's why I'm excited to talk to Devin, because this use case, we are all customers. The United States Postal Service operates one of the largest customer service operations in the world. Every address, every community, every business, So Devin, let's get into it.

Awesome.

When you first launched Agent Forest last November, what surprised you the most once you moved from concept to operations?

Awesome. So that's an excellent question, Mia, and thank you. That's a great starting point. So when I think about that, what we did foundationally right last year is that we made a deliberate investment in implementing a strong governance and risk tolerance model long before our deployment. So what we saw from that is that created guardrails for us to be able to scale when we would eventually deploy AgentForce very responsibly and confidently without hesitation. So when we ultimately deployed AgentForce last November, right before peak season, might I add, wasn't a conscious choice. And so what we saw from that and what we predicted was this normal user adoption arc, right? So when you think about it, awareness, evaluation, trial, adoption, and gradual advocacy. But what was most surprising in that scenario was how quickly our user base evolved from phase 1 awareness to this full shift to full-out advocacy.

Wow.

And so it was in that acceleration that we learned that this wasn't just another tool or a better-built mousetrap. This was a core fundamental shift in our operating model. What was interestingly and even more surprising was that conversations with our senior leadership almost took that same trajectory and path. So questions rapidly evolved from what is this technology and what does it do to how quickly can we scale it and deploy faster. And so those were key unlocks. What we recognized is we were yielding from the benefit of setting governance right upstream and that we were able to get that right, where we were able to scale confidently, we'll be able to— we were able to understand our public and realize our public trust mandate and further our personal mission, universal service of mail and package delivery to every American public. So what was key from that was that what we learned was that it wasn't about, you know, the technology, it was about how quickly our organization evolved and embraced agents as part of our core operating model.

Well, I like that. That's a great segue into what was the misconception, the biggest misconception about the technology before you deployed, and how did that change once you were live? You said, you know, your leadership leaned in and they stopped asking about the technology, but what was the biggest misconception?

I'll tell you, this was a very challenging one. One of the biggest misconceptions upfront was recognizing or thinking that deploying agentic AI would be relatively turnkey.

Mm-hmm.

Once connected to knowledge articles and knowledge systems, knowledge bases, everywhere in between, that the challenge would be technical execution. Or so I thought. What I quickly learned was that the challenge wasn't within technical execution. That wasn't the constraint. It's what I refer to as a lack of agent orientation.

Ah, okay.

So when we think about agents as part of our workforce, albeit digital, then like any new employee, they have to understand how to show up on day one.

Mm-hmm.

So that means agents have to understand how our organization speaks, how we think, and most importantly, how product domains operate. So for us specifically at the United States Postal Service, that meant grounding our AI agents in what I refer to as our institutional language of our organization. Investing in prompt engineering and design and tuning so that our agents can navigate our complex distinctions, domestic versus international, retail versus commercial, how to be able to make sure that our agents can provide accurate and contextually aware recommendations and responses. And so the piece that was not the misconception wasn't technology, it was us thinking that governance guardrails and systems integration alone were enough.

Sure.

And so the key piece from that and the lesson that was clear is that agents in our digital workforce need onboarding too. And so regardless of the training and orientation, if we provide that for our human agents, we have to make sure that we provide that for our digital agents so they can scale effectively and confidently across organizations.

Yeah, especially in government when there's so many acronyms that have the same acronyms that mean different things. All right. So you've learned a lot. So tell us, what do you think is next for U.S. Postal Service or across the federal landscape in general?

That's an excellent question. What I think has evolved so much almost on a daily and weekly basis when we think about deploying agentic AI, one of the key things is, you know, what's next is moving from this assistive AI technology to almost this fully agentic AI solution where Agentic AI is serving as this workforce multiplier of sorts. So it's moving from agent-assisting work to having agent, agentic AI owning the entire work stream. And so in my view, that's not linear, right? And so in fact, with a lot of my conversations with my federal sector counterparts, some of which are in this room, it felt like AI adoption was this arms race, right? There was this push to stockpile and deploy as many AI agents as fiscally possible, and that's not a winning strategy.

No.

What is a winning strategy, and organizations and agencies who will move fastest in this area understand that it's not about the most AI, it's about how well those components are trusted, governed, and deeply integrated in mission-critical systems that enables us to scale. And so when I think about this, Leaders in this area will get 3 things right. This is super critical. They'll govern before they design. They'll think about and harness agents as part of their workforce and their digital workforce, not just as AI assistants and automation, and that they'll design for intentional orchestration across all experiences where agents are able to seamlessly and confidently interact across systems and technology to be able to provide an ultimate result for our organizations and agencies.

I love that. Now, that's what Mission Speed will look like in action, ladies and gentlemen. So, Devin, thank you so much. Look, we got to keep that same energy as I welcome two of my favorite colleagues to the stage, Mia Rodriguez and Nadia Hansen. And around here, we call ourselves M&M. Now, listen, we can't leave this moment. We've got Devin here. We've got to get a quick selfie, ladies. Do you agree?

Yes, we do.

Let's do it.

I'm going to have to bend down.

Yeah, you're going to have to. All right.

Bear with us, please. Turn it around. All right.

Perfect.

All right.

Well, let us get out of here, Devin, before Nadia turns us into an AI agent or something.

Well, thank you, Mia. That was a lot of fun. Thank you, Devin, for participating in our selfie. I'm Mia. I'm the other Mia. And I'm here with my friend Nadia. And together we are going to talk about how our products and solutions work together. Now, before we tell you what's hot in 2026, we're going to talk about some of the trends that overtook us in 2025. And I need your help. So raise your hand if you heard about the Labubu trend. Anybody? Oh, that was a very unfortunate one. How about the Six Seven trend? Yeah. And then the Dubai Chocolate trend. Well, congratulations. We are all trendy. Our children will now finally be proud of us. Nadia?

Hey, Mia, we can't say 6-7 and not do this. You guys, my kids are in the audience, so I have to be as cringe as possible. But in all seriousness, what's not a trend is everyone in this room is being asked about AI. Your leadership wants it. Your constituents expect it. Every vendor pitching you is selling it. Now, I had the privilege of serving in government for 15 years, and to be real honest, technology was never the problem. It was always the data. Data being trapped in 300 to 400 systems. Now, in this era of AI and automation, data alone is not enough. Context is the new currency. Context is what provides your data meaning, connects the systems. It powers intelligence across the entire enterprise. Without context, your AI agents, they don't know how to act. They wouldn't know how to advise. They don't know how to react. So think of data and context as your new infrastructure for AI. Now, speaking of agents, here's what I love about AgentForce Public Sector. The best part is these— this is not a blank canvas. Prebuilt configured agents that really understand how government works, whether it is case management, benefits eligibility, licensing, permitting, inspections, grants management, economic development, you name it. So the best part about them is this is beyond your static chatbots that are, by the way, so 1998. They provide relevancy, have a conversation, and they are live in production in agencies like the USPS right now?

Absolutely, Nadia. Agents are very different from chatbots. Agents have the power of deductive reasoning, so they are providing us with more than just static answers. They have guardrails that we can scale up or that we can scale down. Now, while chatbots are more reactive, agents are more proactive, so they have the ability to solve more complex problems with minimal human intervention. In fact, it's these agents that power Agentforce Public Sector, which is our AI-driven platform that we've created specifically for you, specifically customized for public sector agencies so you all can work on some of your most complex problems. Now, let us imagine that we are in an orchestra and many of you played band in high school, so you probably know what an orchestra sounds like. Now, in the center of our orchestra is where our flutes and oboes play. That's where our data is. Good data, clean data grounds you in the full picture. Good data ensures that we make less errors, we have full decision-making power, and we have more accountability. Good data ensures that we have reliable outcomes, reliable outcomes for us, for our teams, and for all the communities that we serve. Now, let's stay with our orchestra theme for a second. And if data is the center of our orchestra, HEnforce Public Sector is the string section. The conductor of this entire orchestra is Slack. Slack is truly the center of effective collaboration. Your teammates, your agencies, your contractors, all of you are speaking in the secure environment. You have access to workflows, recaps, all your history, all your conversation accessible all the time. Slack is truly what makes this orchestra make more beautiful music. So then if you need a higher degree of compliance, we have an instance of Slack, GovSlack. GovSlack is both FedRAMP High Authorized as well as IL4 Authorized. So you can be rest assured that all of your conversations are happening in a super trusted environment. Now to tell us more about trust, It is my pleasure to welcome my friend Karen on stage. Hey, Karen.

Welcome.

Thank you so much, Nadia. Thank you so much, Mia. I'm so excited to be here with you today. I'm excited to hear how we're bringing data, my love language, apps, my playmakers, and agents, my secret weapon, all into the flow of work with Slack. This creates a winning team on a platform that is mission ready for any public sector use case. As a product manager, I live for the moment to be able to bring the right products and capabilities to you, to the right people. And as we know, with March Madness, we know that every second on that clock matters and that you need answers right now. So let's move at mission speed. Today, I'm excited to announce our new industry innovations and compliance achievements. Coming this summer, we have AgentForce that will achieve IL5 on the most secure defense environment. Next, we had field service and operations bringing you unified scheduling and offline mobile capabilities. And finally, we have Agentforce IT Service for public sector providing you AI agents that independently handle IT requests. So what does this all mean in practice? Well, let's see a demo of Agentforce IT Service for public sector and how it takes the manual out of maintenance. Let's meet Julia. It's her first day as a new state agency employee. She's usually flooded in paper forms and waiting days for basic equipment, and this delays her ability to serve new constituents. But now we can do things differently. Please welcome my demo driver, Matt Atkins, to show you how.

Okay.

Julia opens her Slack IT support channel. And she asks IT— Agent Forest's IT employee agent about getting a new monitor. Instead of a generic link, the agent reasons through the request, instantly verifies her role, and then checks the configuration management database for her location and shipping address. It also checks the inventory to see what is available to her. The agent now has independently processed the request. The agent asks if there's anything else that Julia needs help with. Julia is unable to connect to the VPN. The agent brings back common solutions from a knowledge article, which saves Julia time from having to search her onboarding documents. Now the agent understands that this is a Tier 2 technical issue requiring more advanced technical support. The agent offers to create and submit an incident on Julia's behalf. While the incident is in progress, Julia continues her onboarding. Now let's join Mark, an IT specialist who's just starting his day. He immediately sees the incident in his Agentic IT service desk. And with Agent Assist on the right-hand side, he's able to see a number of employees that are able to also having VPN failures. He checks the detail, he validates the pattern and confirms there's a number of issues across the agency. With Mission Speed, Mark is able to promote this to a major incident so it can be prioritized. Now the incident is in progress. Now let's fast forward a half an hour later. An IT infrastructure teammate was able to pinpoint the issue. With this information, Mark updates the feed, and now with the help of the AI resolution summary, he's able to wrap up the incident promptly. Now, what you just saw in the demo was a single entry point for Julia to connect to multiple subject matter agents to help with things like VPN troubleshooting and hardware management. Mission delivered at mission speed. Thank you, Matt, for that wonderful demo. Agent Forest helps IT departments scale and be efficient without adding headcount while reducing resolution time. This is the definition of efficiency. But we know in public sector that when you give the agents the power to now access system logs, well, now the conversation changes to governance. With public sector, Efficiency means nothing without the certainty of trust and the mission-grade security of Mission Force. Let's check out this short film.

Somewhere out there, a mission is underway. We may never know what it is. We may never hear about the outcome. But we do know this. Every mission begins with purpose. Behind that purpose are people. Not just any people. The most extraordinary among us. The ones who run into the fire, who hold the line. The allies who guard the peace. The people who put their lives on the line so the rest of us can live ours. So our dreams can endure. But too often their missions are slowed by silos, buried in outdated systems, trapped by processes built for another era. What if their courage to achieve was matched by their ability to achieve it? What if their sacrifice was met with systems that never failed them? What if their resolve was met with the clarity to act without hesitation? Not just another parade or a thank you, but a commitment. A commitment to the most extraordinary among us. We call it Mission Force. Mission Force is more than technology. It's trust in the moments when decisions matter most. It's acceleration when there's no time to waste. It's breakthrough innovation when bureaucracies and the status quo put lives on the line. Because missions aren't won in silos, no. They're won in the fire, in the field, in the quiet decisions that no one ever sees or hears. They're won by people. Mission Force is built for them. For the defenders. For the loved ones who wait. For the allies who stand beside us. It extends the best of the private sector to the public sector, to their families, and to our veterans. Equipping our nation's defenders with the tools they need. Powered by agents and data, driven by Automation App, united through collaboration, built with urgency and agility, with the scale to deliver meaningful taxpayer savings, all while advancing the readiness of the United States and our allies around the world. This isn't just modernization, no, it's mobilization. A new way to serve those who serve us all. Because when the people behind the mission rise, we all do.

In the mission for the U.S. Army Human Resource Command, is to serve the 9.2 million soldiers, veterans, and civilians through their Army Service Center. To tell us more about their modernization journey to a digital front door, please help me welcome to the stage Colonel Christopher Brown.

Hey, Susan, I'm not sure how I follow that.

No, it's inspiring, right? But so is the work that you are doing. And we were talking earlier, but just about the groundbreaking progress and the thinking that's going into the work. So to not delay it any further, please share with the audience a bit around your journey to the digital front door and also just like what you were able to unlock for your teams in terms of mission readiness and scale.

Absolutely. So as you can imagine, you know, as the video highlighted, the people, right? That is our business. And I've said for years, it can be a messy, business, but it's something we can't fail on. We have to do it right. And through Salesforce and this digital front door that we're building and implementing, we're going to get it right this time. So we're going to deliver solutions to our customers. As you said, that's over a million service members that wear this uniform, almost another 10 million veterans that used to wear that. Plus their families. And so that's a critical need to get it right. And so that means we've got to take the new technology and be willing to replace our old legacy systems with that. And so what we're doing is opening up that digital front door. Initially, it's going to be in-house, um, but to a call center that those 10+ million veterans and service members can reach out to to get their questions answered. Today they have to send an email, they have to pick up the phone, and they have to deal with a customer service agent. That won't go away. As you said earlier, that we have to keep up that base and that partnership of the human and the technology. And so we're going to continue to do that, but we're going to integrate a digital workforce. And so that will allow us to now pull off some of the jobs that are a little more mundane and allow that digital workforce to take care of that while the humans take the more complex problems and work those, right? Those are the things that need human touch. And so that's our vision. That's how we're implementing the digital front door and starting right now to change the way Human Resources Command does business.

Well, that's great. I mean, just describing is exactly the ideal scenario, right? Is that partnership with humans and agents. And in implementing this project, you guys have been able to achieve tremendous results. You finished this 4 months ahead of schedule and under budget, which we don't often see. So can you share with us a bit about your approach and how you're able to achieve those results, but also like the value that you've seen so far?

Absolutely. So This is a balance, right? When it comes to implementing a new system, we can't do it by ourselves. We called on Salesforce to come in and help us, right? And through that partnership and other partnerships, right? Like listening to the USPS and what they went through, learning through other agencies, we've changed the way we do business. It's no surprise to you that the government is a little bureaucratic and maybe slows things down. And it usually is faulted on risk, right? Nobody wants to take risk. So we've taken a different approach. We've looked at industry and said, let's do more of what they do. Let's take that agile approach and that fail fast mentality. And so I've challenged my entire workforce to do exactly that, take risk and fail fast. And they know what their boundaries are and where they can or can't take risk and when it needs to come up to my level. But through that process, that new mindset, the partnerships that we have and the ability to pull in new technology, it has really allowed us to accelerate where we're at. The other piece is, as you heard with business, business processes, the willingness to change business processes and harness the technology as it exists. So again, we like to overcomplicate things. We like to customize things and do it our way. Well, when we cannot do that and we can use something that's more off the shelf, exponential growth. And so that's one of those amazing things that I've seen through this process with our partners here. To really move us ahead and honestly win me a few favors with the boss.

That's great. And, you know, and thinking about that, right, and being able to even change the way that you implement and you look at it, I think in this age of agentic enterprise and AI, we have to. We can't implement the software from 20 years ago the way and apply those principles to the way we're operating today. So as you think to the future, And what does the agentic future hold for your digital front door?

All right, so I told you earlier, I'm a career human resources, so I'm not an IT guy. I'm the guy that's been out in the field doing the work. So I've been imagining this for 20-plus years. How would I do it different and do it better? And this is it. This is what we're going to do. We're going to implement agents that from the outside, right? So as a customer, I reach out, I start a chat and it says, okay, Chris Brown, glad to talk to you today. I see that you are overdue on your casualty notification documentation. Do you want to update it? Well, that wasn't what I was here for, so no. It'll say, okay, I'll email you and remind you that you need to do it. How simple is that, right? Now, let's go internal to the system and let's change the way that we're processing things. And so we're implementing agents. So what used to take weeks and months to process because this person had to look at it for this piece of data, this person, and right, it went through this long process. Now we have an agent or agents, let me rephrase that, that can touch all those pieces And I'm not changing who sees it in the end, right? That person's still sitting there to be the final validation. But I got a green checkbox that said it's already been checked and ready to go forward. So I just, again, cut out time, which gives that time back to the soldiers and service members that are out there because they're not waiting to hear back from us. They're not waiting 6 months to get feedback on, can I retire or not? We're doing it in weeks now. And then you take it all the way out to the commander's level, and we're providing commanders in the field predictive notifications of when soldiers are going to move and what their units and strengths are going to look like. And so now they can make a decision on who is going to be available for a rotation to somewhere overseas or maybe a training exercise. Where they didn't have that complete predictability before. And so really across the entire enterprise, we're looking at how we can implement agents to simplify, augment, and support our teams so that we really can move at mission speed.

Great. Well, thank you so much for being here with us today.

Absolutely.

Thank you for sharing that.