Workday CIO and CTO on Why the Agent Era Belongs to Platforms

I'm not willing to risk my job. I'm not willing to risk the trust of our internal stakeholders. I'm certainly not willing to risk the trust of our customers. When you're dealing with something that has a high trust requirement, you just have to deal with agents that you trust. That was Ronnie Johnson, Workday's Chief Information Officer. And she's not talking about some hypothetical. She's talking about the very real pressure CIOs face right now. Build AI agents fast or fall behind. But move too fast and you could break the systems your business runs on. In this episode of the Future of Work podcast, Ronnie and Workday CTO Gabe Munroy break down the new playbook for agents, from reorganizing your teams to knowing exactly when to hit the brakes. Let's hear from them both. We may be at a developer conference today, But here we're going to talk about the people who own the outcomes, the leaders. I'm Ryan Basilio, the Director and Head of Outbound for the Developer Platform, and I'm joined by Workday's CTO, Gabe Monroy, and Workday's CIO, Ronnie Johnson. Guys, welcome to DevCon. Thank you. Yeah. So we just heard some amazing announcements from the keynote. Any initial reactions to the developer agent or anything you saw? Well, what I could say is the reactions from the audience and the folks that I've talked to have been, I think on the one hand, just massive excitement around how amazing the innovation is. And on the other hand, there's a lot of reflection going on around, you know, people thinking about what does this mean for me? What does this mean for the way that I worked in the past? Because this agentic tech, it does a lot. Yeah, yeah, it's gonna do a lot. And I know we put a lot into it, but I wanna bring this back to the way we're thinking about it from the industry writ large. Ronnie, the playbook, so to say, for how to implement AI still feels like it's being written. How ready do you think we are as an industry and what are you seeing from your peers and some of the customers in the space? I think there's really two sides to this. One is, is the software really ready to deliver the value, you know, or the hype that we've all talked about? But the other part is, are our IT teams, are our technology and delivery teams, are they actually ready? Do they have the skill sets and the operating models to deliver on the value, to actually deliver the efficiency gains and the kind of productivity gains that are expected by that? I'm reminded of my kind of experience in the first 3 years that I've been delivering agents. One of the things that was interesting is in our operating model, most of our dev teams sat in our data science teams and our infrastructure teams, and they didn't have proximity to the business or the intimacy necessary to actually meet with our business partners, determine how to actually deliver value, and figure out what were the most important and high-priority use cases. And so we've had to do a lot to reorganize how we do business engagement but also had to reorganize our teams to focus on how we deliver. We've moved from the kind of traditional agile pods to what we're calling AI pods now. And we're seeing more kind of real full-stack engineers in our delivery models versus where we were doing the traditional handoffs between BSA to dev to QA and to delivery, our DevOps teams. And so we've really had to kind of reorganize to make sure we can actually get to delivery. So you're right, playbooks are still being made, being invented, but I am, I do believe the hype. I'm feeling the hype and we're seeing, you know, seeing value in the hype. You mentioned a word specifically, and I want to ask Gabe if he's seen this in his past life. You mentioned the intimacy that the team needs to have in order for these to be successful. Gabe, how did you see that work out in your previous roles and the other organizations you were a part of before? Can you comment on the necessary needs for intimacy, for closeness to the business? Define intimacy more. Like, what do you mean by intimacy? Yeah, Ronnie, I know you were talking about— So for us, what we'd had to do is go to each one of our business departments and actually curate use cases. But if we, you know, let's say it's your infra team who's, you know, that's focused on really delivering integrations or network, they didn't actually understand what was valuable to the business. They didn't understand the business outcomes that were important. And so that intimacy of understanding what were the important business outcomes to figure out how to curate the right use cases and how to prioritize where to deliver AI. I think that is so important because the premium on use cases has gone up tremendously in the world of AI and use case enumeration and proximity to the business. In a weird way, it's almost like 12, 18 months ago, a lot of these initiatives were blocked on, do I have the skill sets? Do I have the resources? Where can I fit it into my roadmap? And what we're seeing with some of the news that were coming out, Developer Agent being a good example, is like you can now build at the speed of thought, at least in a lot of cases, not every case, but a lot of cases. And so now the questions come back to the business around, Well, where do we want to prioritize investing? And that proximity and that intimacy that you're describing is critical. Yeah. No, I love that. I love that. And you mentioned kind of like the hard part about it, and you were alluding to it at least. Gabe, what are the considerations for leaders on how or where to build, right? It's kind of like you were talking about what to build, but how or where? Do you have an opinion on that? Should we have an opinion on that? Can you dive a little? Yeah, I think it's important to understand what these AI systems are good at and what they're not. And what they're great at is areas where creativity and reasoning and open-ended solutions are helpful. That's why drafting an email or a document or helping with creative writing is such a common task for AI. Where things are trickier and a little more nuanced is where you really need deterministic guarantees, right? And in the area of people and money, you really need those deterministic guarantees a lot. So we have to tread carefully as Workday, and I'd argue our customers as well, on how they start to apply AI in areas adjacent to people and money where determinism is a must. And you got these AI systems that don't necessarily guarantee that. Now we're working hard to engineer solutions to these problems, but I do advise CIOs to just be careful. Yeah. Ronnie, comment on that because Gabe is saying CIOs be careful. How are you saying this message today to your peer set and to your leaders as well? How are you propagating this? So when you think about the landscape of a CIO, we're managing somewhere between 100 to 500 systems in our overall ecosystem. And so there's times where it's okay to— I'm going to borrow our Chief AI Officer's words— it's okay to have YOLO agents on certain things. And so you only live once. It's okay if you're in a role where it's acceptable to have agents actually responding to your emails automatically because you're doing routine tasks, then YOLO. But when it's time to do something, as you mentioned, that has to be deterministic, where it is, you're doing a mission-critical service that has to absolutely guarantee the right result, where you're talking about writing to a system of record that is providing, you know, either a critical answer that has regulatory controls or frankly is dealing with sensitive data. You really, you can't YOLO that type of work. And so you really do need to be thoughtful in using lawful actions in that. Feels like we've gone the spectrum, right? From like YOLO it to, wow, this is going to be a real problem. Maybe even a serious problem. Maybe someone doesn't get paid. Maybe you break the law unintentionally. How are we thinking about trust or validating trust, right? We can hit the deterministic aspects, but comment for me on how Workday and how you would like to see trust propagate through these agents and maybe for agents in the future? Well, you know, one thing that we announced at the show here is our Agent Passport solution. And this is something that we've worked with Cisco as a launch partner. You know, customers are asking, how do I understand whether this agent, you know, has been tested, has been validated, you know, is going to provide a set of security, governance, and controls? And a lot of customers are looking to Workday to provide those guarantees. And so what the Agent Passport program does is it provides stamps, attestations around security capabilities to prevent things like prompt injection. And we're constantly testing and validating those attestations. In the case of Cisco, we're working with them on their AI Defender technology, you know, which allows, you know, Cisco to sort of test at runtime as well as at release time whether these agents are actually living up to the expectations that CIOs have around security and governance. So that's one concrete thing we're doing now. Actually, Ronnie, I'm curious, your thoughts on that kind of approach. I mean, you can't not not, if that makes sense. I mean, when you're doing something on a critical system with sensitive data, I'm not willing to risk my job. I'm not willing to risk the trust of our internal stakeholders. I'm certainly not willing to risk the trust of our customers. When you're dealing with something that has a high trust requirement, you just have to deal with agents that you trust. One of the things that I think is important, and I think this is common amongst my peers in the CIO industry, is that we really want to have agents operating on the— within the kind of the system of record where they understand the context, they understand the business processes, they understand the controls. And so many of us aren't comfortable enough and probably won't get comfortable enough having an outside third-party agent taking actions into a system of record, especially when you're talking to writing— writing to a system of record where there are some regulatory control requirements. I absolutely love that. I'm going to kind of take us off the rails a little bit because you sparked something interesting. I know we had talked about reasoning before, Gabe. We were kind of over there talking about reasoning and the experience layer. Where does that trust— where do you draw the line on the experience, on the runtime? Tell me a little bit about where you're thinking about what you'd recommend to our customers who share the exact same concern as Ronnie here. Yeah, and we put out a blog post on this. I authored the blog post actually. It's called The Three Paths to Build, and it outlines a spectrum of ways to interact with Workday on an agentic front. The first The first path is really about integrating with the system of record using MCP tools and traditional AI automations, right? So it's like an API loosely coupled model. And as you said, Ronnie, like you gotta be careful with the writes. The reads generally, you know, tend to be okay. The writes tend to have some risk. And so what we've really aligned on is a second path where the reasoning and the LLM sort of execution is happening on the Workday platform because that proximity to the system of record, allows us to provide a level of guardrails that simply are not possible to do if those agents are running in Anthropic or Google or some Microsoft system. We, you know, we love those partners, we work with them deeply, but sometimes you really want that reasoning to happen, you know, inside of Workday and that's how you can be more lawful in your execution. And then the third path is really, you know, moving beyond just the chat interfaces into a full UI. You know, what happens when you need a visualization or a chart or a multidimensional thing that doesn't sit inside of a chat that box nicely. Our bet there is Sana, and Sana is that user interface that allows us to truly reinvent the world of HR and finance in an agentic way. Love that. Love that. You mentioned the chatbot. I would assume you can't do your entire workflow out of a tiny chatbot, and we need something a little bit more. Ronnie, can you corroborate that, or is chat the only way? I wish I could chat my way through my job, but no, chat is not the only way. Okay, just wanted to make sure. Get that one out for the audience's sake. We're running low, but I did want to leave. I want to kind of get some takeaways. Ronnie, what would be the kind of next step for any of the executives listening, any CIOs listening in terms of what to do next? How do they approach this? What, what is your immediate takeaway or step for them to do? I mean, the very first thing is if you haven't started building agents, if you haven't started developing, get started. Agent development is not going to just change the expectations around efficiency and optimization, it's going to actually change your operating models. It's going to require new skill sets of your team. So there's— sitting on the side is not going to get you more ready. Waiting for the industry to mature is not going to get you more ready. You have to get in, start to learn how this works, see how it evolves your organization. And then the wonderful thing, you know, if I can make a plug for Workday, I mean, there's some things that are just not regrettable. When you know you have to manage a system of record, when you can deal with native agents in that system of record, and you can actually work on a trusted platform, it's an— to me, it's a non-regrettable choice. Definitely. Gabe, any last comments for anyone listening out there for what they should be doing next? Well, I agree with Ronnie. This is a moment where the industry needs to lean forward, right? There's a lot of concern about what does this mean for my industry, for, you know, how we've operated in the past. And oftentimes we don't have answers to those. And you're not going to get the answer until you lean forward and start doing it right. And a lot of this is very contextual to the specific company, to specific organization, to the specific function. And you really just have to put one foot in front of the other, start moving forward and start building. I'd really encourage folks concretely to take a spin with our developer agent, right? The ability to go build new innovation in Workday using Claude Code, using Antigravity, using like all these great AI tools. You'd be surprised at what's possible to do with the Workday system now, even without a lot of technical skill. That's really changed. But you got to try it. Got to try it. Got to lean in. That's what I'm hearing. Wanted to say thank you both for the conversation. Thank you everyone who joined us. Be sure to subscribe to the Future of Work podcast and hear more conversations like this one. Gabe, Ronnie, so good to see you. Thank you.