At its annual DevCon conference in Las Vegas on June 2, Workday unveiled three new capabilities that move enterprise AI agent deployment from a trust-and-hope model to a verified, governed system. The announcements — Developer Agent, Agent-Ready Tools, and Agent Passport — target the core friction point that has kept many finance and HR teams hesitant about deploying AI agents: knowing whether the agent is actually safe to run.

The Problem Workday Is Solving

Businesses running AI agents in HR and finance face a question they currently have no reliable way to answer: how do I know this agent won’t leak employee salaries, get confused by a system prompt injection, or silently start doing something it wasn’t supposed to do?

Most enterprise AI deployments today rely on vendor assurances and internal testing that was designed for software, not autonomous agents. Agents are different. They interpret instructions, chain multiple tool calls, and make decisions in context. A standard unit test does not catch prompt injection. A code review does not reveal what happens when an agent is asked to bypass its guardrails.

Workday’s announcement is a direct response to this gap.

Agent Passport: Verification Before and After Deployment

The most consequential announcement is Agent Passport. It tests every AI agent — whether built by Workday, a third-party developer, or a customer’s internal team — before it goes into production, and then monitors it continuously afterwards.

The testing framework maps to public industry standards: OWASP LLM Top 10, NIST AI Risk Management Framework, and MITRE ATLAS. These are the benchmarks that enterprise security and compliance teams already reference. Agent Passport runs agents through scenarios designed to surface prompt injection, jailbreak attempts, goal hijacking, system prompt extraction, leaks of employee data, and generation of unsafe outputs.

Each agent that passes receives a verifiable record of what was tested, what the results were, and what standards the testing covered. When that agent is updated, the Passport updates too — because continuous monitoring runs after deployment, not just before it.

This is a meaningful shift from how most enterprise software manages risk. Software ships with bug reports. Agents now ship with governance records.

Agent Passport will be available to early access customers in the second half of 2026, with general availability projected before the end of the year.

Developer Agent: Building in Plain Language

The second announcement is Developer Agent, which changes how people create AI agents that run on Workday.

Today, building a custom agent that connects to Workday data typically requires developers to understand API schemas, authentication flows, and the platform’s data model. That creates a bottleneck. The teams that understand the business problem — finance analysts who know what budget variance matters, HR managers who know what onboarding steps are missing — are not the same people who can write the integration code.

Developer Agent collapses that gap. A developer can describe what they want in plain language: “Build an agent that alerts finance when a department is trending over budget this quarter.” Developer Agent then selects the right tools from Workday’s library, connects the relevant data and services, pulls the documentation and examples needed, and builds the agent.

Work that previously required days of setup now takes minutes.

Developer Agent integrates with the tools developers already use: Claude Code, Cline, Codex, Cursor, and Google Antigravity. It uses an open standard called AgentSkills, which means agents built this way can be shared, audited, and extended rather than locked in a proprietary format.

Developer Agent and Agent-Ready Tools are available now to early access customers through Workday Extend Professional, with general availability expected in the second half of 2026.

Agent-Ready Tools: MCP Access to HR and Finance Data

The third component is Agent-Ready Tools — hundreds of purpose-built connectors that give AI agents structured access to Workday’s HR and finance data through the Model Context Protocol (MCP).

MCP is becoming the standard interface for AI agents that need to interact with business systems. Workday’s adoption here matters because it means agents built on any MCP-compatible framework can now access Workday data without needing custom integration work.

Agent-Ready Tools are designed to reduce hallucination and latency. Rather than giving an agent a raw database connection and expecting it to figure out the schema, these tools provide structured, business-logic-aware access to the information the agent actually needs. The agent gets the right context in the right format — which makes outputs more reliable and faster.

For finance teams, this means agents can pull real budget data, compare actuals to forecasts, and flag variances without requiring IT to build a custom data pipeline. For HR teams, it means agents can check headcount plans, pull compensation data for benchmark comparisons, or surface compliance gaps in job classifications — all with the access controls Workday already enforces.

What This Means for Business

The pattern Workday is establishing at DevCon is worth paying attention to, because it signals where enterprise AI is headed more broadly.

Governance is becoming table stakes. Six months ago, enterprises were asking whether to deploy AI agents. Now the conversation has shifted to how to deploy them safely. Agent Passport gives compliance teams a framework for answering questions they previously could not. Boards and auditors will start expecting this kind of documentation.

MCP is becoming the enterprise integration standard. Multiple major platforms — Microsoft, Workday, Google — are converging on MCP as the way AI agents access business systems. Businesses that build their AI agent strategy around MCP-compatible tools will have more flexibility and lower switching costs as the ecosystem matures.

Natural language agent creation is now real. Developer Agent is not a demo. It is designed for the developers already on the platform, using tools they already use. The implication for enterprise teams: the people closest to the business problem can now build agents that solve it, rather than waiting in a development queue.

HR and finance are the next major AI agent domains. Voice AI and customer service agents got there first. Workday’s announcement signals that the back-office functions — payroll, forecasting, compliance, onboarding — are next. Companies that wait for this category to mature before planning will be behind the businesses already running pilots.

---

Building an AI agent workforce across your business operations requires both capability and governance. Enterprise DNA’s Omni Advisory service helps leadership teams design agent deployment strategies that your compliance team can actually stand behind.