Workday AI leaders Dean Arnold and Jay Wieczorkowski reveal how agentic development is finally eliminating the compromise between rapid software deployment and enterprise-grade trust.
Ghadeer Redler
Head of Multimedia, Thought Leadership & Customer Stories
Workday
Payroll can't be 99% right. Neither can benefit enrollment, or a compliance filing, or anything else running through the financial and HR systems that keep a company operating. That's the bar agentic AI has to clear in the enterprise: even as agents reason, adapt, and act with more autonomy, HR and finance leave no room for "mostly right."
It's the exact tension Workday's VP of AI platform Dean Arnold and GM of developer platform Jay Wieczorkowski tackle in this episode of Future of Work. Their answer reframes what it means to build responsibly at speed: give agents the context to reason well, the guardrails to act safely, and a way for CIOs to verify that trust before an agent ever touches production.
Listen to the full conversation to hear how Workday technology leaders are closing the gap between agentic speed and enterprise-grade trust, and what it means for developers building the next generation of workplace AI.
Host: What if you could build and deploy production-ready enterprise software in minutes—without making your CIO break out in a cold sweat over security? On this episode of the Future of Work recorded at DevCon 2026, Workday’s Vice President of AI Platform, Dean Arnold, and GM of Developer Platform, Jay Wieczorkowski, reveal how Workday is eliminating this historic compromise between speed and trust. Let’s hear from Dean and Jay.
Jay Wieczorkowski: It's great to be with you.
Dean Arnold: So, we're going to jump in here. Got some questions for you, Jay. So, Developer Agent.
Jay: Yes.
Dean: How do you think developers react to that? How's this going to change what they're doing in the platform and how they're building?
Jay: Yeah, agentic AI is not incremental change for developers, Dean. It's completely rewritten the entire playbook. You know, it used to be a focus about writing code, and now it's focusing on scaling impact. And with Developer Agent, we're meeting that moment. Developer Agent is going to allow the Workday developer community to build applications and agents that run on Workday from their agentic tool of choice—whether that's Claude, or Cline, or Cursor, Google Anti-Gravity, their own tool they've built. They can plug Developer Agent in via MCP. And with plain language and prompting, they're building across Workday. There's 50-plus skills baked into Developer Agent. All the best practices of how to build right into Workday are baked in. And so now developers are building at the speed of their own thoughts, and the pace in which they're able to build and deliver, but also the impact that they're able to scale themselves to is transformative. We showed it live in action today, and we have people behind us using Developer Agent literally right now for the first time building agents that run on Workday. And it's stunning developers. They're realizing the power that was just put into their hands, and I think they're already thinking about how are they going to bring that back to their organizations. So, we're really excited about Developer Agent. I think it's a game changer.
Dean: Yeah, it's a game changer. I mean, the other interesting thing I find about it is we're also changing the accessibility of who can actually use it as an agent, because now you can just have the conversation. You can just engage that way. You don't have to be in the platform and know everything. Just gets you started so fast.
Jay: Right, and it fits into your total agentic development workflow. So, we recognize that when developers are building on Workday, they're also building with other frameworks and technologies and tools. And sometimes they're bringing Workday data through our APIs into other stacks. And Developer Agent fits into that total orchestration of development, and it helps them skill up at the same time at the pace of their own thoughts. It's wild, it's so cool. We're having a great time here using it for the first time.
Dean: One thing that's important for the developers is they're building agents, and they're building it for our customers' HCM, they're building for finance. How are we approaching powering the agents to act? How are they engaging, and how are they interacting with the platform? What are we doing different?
Jay: Yeah, I think it's a very difficult space when you think about HR and financials—payroll, benefits, the ledger. Things have to be right. And so, we've spent a lot of years, and our customers certainly have, and our developer community has, building what I call enterprise truth into the system. And that's company rules, and data, and context, and business processes, and security policies. All of that power is how the world works. And it doesn't go away in the world of agentic AI; it's actually more necessary than ever because that determinism is what grounds agents. It's fuel. Agents are probabilistic. They reason a lot like you and I. Their context can get overloaded, and they can make mistakes or do things wrong, or hallucinate—that's a fun word.
Dean: Just like us.
Jay: Just like us! And in the space of HR and finance, that creates real impact. You're talking about compliance issues, you're talking about people not getting paid, you're talking about people not having benefits set up for their children.
Dean: Payroll can't be 99% right.
Jay: It can't be! It can't be. And the Workday developer community knows that and takes that incredibly seriously, and so do we. And so with Workday Build and what we're doing to make it agent-ready is all about meeting that moment of keeping enterprise truth grounded in Workday, and you keep layering that into the system. It doesn't stop. You don't stop building new objects and processes and security policies. But then also, how do you build agents that are running on top of that within the guardrails? And so what we did is we built a new suite of purpose-built AI APIs and connectors to power agents. We call them Agent-Ready Tools, and we announced that yesterday. And Agent-Ready Tools are specifically designed to power agents and optimize their outcomes. We found over the last few months that when you take APIs that were built for another purpose, like integrating systems, that required humans to build ETL on top, it's not the right grounding for agents. And so we reimagined how we approached giving agents the right context and being concise, intelligent, and cross-referencing to them so that they can reason effectively. And so we have hundreds of Agent-Ready Tools here that are exposed over MCP. You can bring them into your agents whether they're running on Workday or anywhere else. And we've also unlocked all of the 1,000-plus Pipe Dream connectors as actions that can be embedded into the agents as well. And so as you're building the agents that interact with Workday, you're binding these Agent-Ready Tools to them, and they're able to start executing and acting across the system. And so it's really wild. And again, we have thousands of developers building all of that for the first time behind us right now.
Dean: Yeah, it's incredible. And I just love the way not only we delivering the Agent-Ready Tools, but the agent actions, but you can also combine them together and create your own deterministic workflows as well, because that's important for developers as well. It's not just the ones Workday ships, but the ones that they can also create themselves.
Jay: Yeah, there's times where the agents need to reason and decide something, and then there's times where they just need to act, and you need to tell them what to go do. And so we're providing a suite of purpose-built tools to enable agents to do that over MCP. So, Dean, I got a question for you. You asked me about how we build the agents, we talked a lot about how you power the agents. You announced Agent Passport yesterday. And I'm personally really excited about Agent Passport because it gives developers the ability to stamp and verify that their agents are compliant with open standards. And we've brought—I know you've done a lot and your team, you and your team have done a lot—to kind of bring that to the forefront and bring independent verification and layer that on top of Workday. Could you maybe talk a little bit about that, and how you think that changes the game for CIOs as theyre kind of figuring out how to deploy the agents across the enterprise and making sure they're complying with everything they need from a regulatory and industry perspective?
Dean: Yeah, no, absolutely. So, I think we think a passport kind of takes the guesswork out of being looking at compliant, looking at vulnerabilities, knowing that you can trust the agents that you're either building or people are delivering for you—so the ones you use in your enterprise. And what we observed is customers looking at making decisions, this kind of takes a lot of that labor-intensive, continually monitoring, staying up to date on regulations—it takes that out. And certainly when you're talking about regional compliance like EU AI Act and stuff like that, that's really hard for customers to know all the ins and outs of how these regulations change all the time, who's doing what, what do I need to change. And having a solution that makes that much simpler for customers enables them to remain compliant.
And something we did that's kind of interesting with passport, I think it's very differentiated in the industry, is we took this approach where we created these things called stamps—passport stamps, who knew! And these stamps kind of bring together all these different industry standards together, and we create these categories of stamps, and we can group them together so they're easy to understand. But it also means we can layer up lots of different third parties, different attestation partners. And we announced Cisco yesterday as our first leading partner there. But we can create these stamps and we can have different partners participate in them, and it allows you to create trust from lots of different partners and say, "Wow, this agent uses this partner, and we can attest over there; and we're using this partner to do the test over here with different stamps." But the stamps create that taxonomy. It means that you have a standard way of looking across different agents across different regions and understand that very simply: I have this stamp, I have this stamp. And with Agent Passport, I think we're releasing probably about 30 different stamps when we actually release it later this year. And we have a good number going out the door now that we're trialing now. So, I think it's really interesting. It's very different to what we've seen from other vendors in the market as well, and we're really excited about it, like you said.
Jay: Yeah, I love the transparency of it, Dean. You know, we've talked about it that, obviously, Workday is going through a large amount of effort to bake that trust into the agents that we're delivering and provide it through our runtime. But there are also regulations and compliance and standards that are specific to industries or geographies, and it's all emerging, right? And so, you know, I love your approach to kind of independently operate with the leaders in the ecosystem. And then what's really cool is you guys are surfacing agent trust profiles right inside the agent system of record. So, it's completely in front of the administrator as they're activating these agents.
Dean: That's right. We're putting trust on the dashboard. It's front and center, you can see it. And as you say, all the different—we have different attestors, different regions, and you can put them in different profiles. So, you can layer them all up for your organization. You can say, "I need this attestor for this region, and this governance and this regulation." So, yeah, it's really powerful.
Jay: And we're going to have developer—we already have developers behind us stamping their agents right now.
Dean: Yeah, taking their agents, yeah, yeah. We spoke about yesterday, we spoke about surface as well. That's the other thing we spoke about. And that's—it's important for customers that they have choice in kind of where they surface their agents, and how their employees engage with them, because we want the employees to engage where they do their work. So, how are we thinking about surface, what are the options, how does a customer make a decision around where they surface their agents and what platforms they should use?
Jay: Yeah, you know, Dean, we've talked a lot about this. The agentic future is open. Open doesn't mean lawless, right? So, you know, when we think about it, we've built three very distinct paths to build that are all, you know, can be used to power agents. And you kind of think about it on two different axes, right? So, on the first, it's where the user's working, and on the other side is who's doing the reasoning. So, you know, we talked a little bit about Agent-Ready Tools earlier, and Agent-Ready Tools are these purpose-built AI APIs and connectors that use over MCP. And those are designed—you can bring those into an agent wherever it's running, right? So, you know, the agent could be reasoning with an LLM hosted somewhere else. We bring the power to act inside Workday, but it is on the developer and that provider to ensure that the reasoning on top of that, you know, is happening intelligently. And that does put a lot of burden on the developer, but there's a lot of scenarios where that may be makes sense, right?
The second path is when you start talking about being able to run the agents inside the guardrails of Workday, inside our platform, and then connect them across the enterprise. And so we recognize that, you know, users are working everywhere. A lot of times they are working in Workday, they're working in Sana—surprise! They might be working in Gemini Enterprise, they might be working in Copilot, they might be working anywhere across our range of partners.
Dean: I think in Workday we have at least three surfaces that we would do that through as well.
Jay: Right! And so developers need the ability to bring what they've built where the users are getting the work done. And so what, you know—I'm speaking on behalf of you and your team, Dean—what you guys have done a great job of building a number of connectivity options using open standards that bring that to those surfaces while keeping that reasoning inside Workday, where we can make sure it operates lawfully. We talked so much about how HR and finance, it's a really important space to make sure things are getting right. And when we run the reasoning inside Workday, you know, we're able to kind of lawfully govern that inside the runtime. And so you've been embracing A2A, MCP, A2UI. So, all of those standards and connectivity options are available to developers to surface their agents and scale them up across the enterprise wherever the users are working.
Dean: That's right.
Jay: You know, and then our third option is we talk about the complete package, right? So, you know, when we want to provide—our developers want to provide a complete end-to-end experience that's curated from the top down, from the experience and the orchestration into the agent, into the acting, and running that through. And that's where we bring Sana into the table, right? And enabling our developers to surface the agents in Sana, you get a complete, again, end-to-end experience for the user. They're logging into the system, the agents are running completely natively inside the whole stack, and you get the full package, right? So, yeah, it's about providing options, it's about being open, but it's about making sure that, you know, folks understand those two axes again. It's about where is it running, where's the reasoning happening, and then choosing the right path depending on the use case.
Dean: And that's really, it's really important, really valuable to our customers. We're providing these different interaction patterns, they have these options. Reasoning stays in Workday as much as you can, that's higher value to them, that's there's reuse there, they can use the agents we provide. And if you like surface inside Sana, as you say, you got full package end-to-end. That's great! Awesome, well, thanks, Jay. Good to catch up with you.
Jay: Always, Dean.
Dean: Thank you, everybody, for joining us. Please subscribe to our DevTalk podcast if you want to hear more conversations like this one.
Report
A gap between effort and impact is the real ROI problem. Discover the solution.