Introducing Workday's Trusted Agent Platform

How it all comes together:

• Lawless vs. Lawful Headless

• Open by Design: Build Anywhere, Run on Workday's Rails

• Backed by Builders: The Open Developer Platform

• Lawful by Default: The Harness Under Every Agent

• What Lawful Context Enables

• Go Build

The rush to go headless can create a wave of lawless agents across the enterprise. Today at Workday Rising, we're announcing Workday’s Trusted Agent Platform, giving teams lawful, headless access to HR and finance data—using whatever tools, models, and frameworks they choose. Every core capability is generally available (GA) today.

When enterprise software touches people and money, speed and guardrails decay with every hop away from your live enterprise state. That’s why no matter who builds the agent, every action running through our platform stays bound to Workday’s permissions, approvals, and audit trails. No agent gets a side door.

Even our own internal Workday agents run on these exact same rails. Here’s how it all comes together.

Lawless vs. Lawful Headless

Enterprise software is going headless, and AI agents are actively executing workflows across HR and finance without anyone logging in. But going headless without native governance creates systemic risk.

Most teams fall into two architectural patterns that quietly break compliance:

• The Custom MCP Pattern: Running a custom server on a single, over-permissioned service account turns the agent into a "god-mode" admin—bypassing approvals, ignoring regional scoping, and destroying auditability.

• The Data Lake Pattern: Copying Workday data to an external lake moves the raw rows over, but leaves your live business rules, organizational hierarchy, and security policies behind—giving agents overprivileged access.

When an agent touches the general ledger, executes payroll, or approves a compensation change, "usually right" brings an auditor knocking at your door. Headless must be lawful. Security context and business rules have to travel with every single request.

Open by Design: Build Anywhere, Run on Workday's Rails

To make headless lawful, we’re opening access to live Workday data, context, and actions while enforcing your security model on every request at runtime.

Two core capabilities make that possible, both generally available today:

1. Agent-Ready Tools (Model Context Protocol): Native Model Context Protocol (MCP) tools that bring Workday’s permissions, approvals, and audit trails straight to any MCP-compliant agent (Claude, Copilot, or custom stacks). Built on open MCP standards alongside Anthropic—a key partner and active Workday customer.

2. Workday Data Cloud with Live Data Query: Direct SQL access to live Workday business objects with zero data duplication, stale copies, or security drift.

Backed by Builders: The Open Developer Platform

Building on Workday should feel as natural as building anywhere else. We completely revamped our developer experience to remove friction without compromising trust. Here’s what’s new:

• Open Developer Platform: Anyone can sign up at developer.workday.com to inspect API specs and browse our MCP catalog with zero gatekeeping.

• Developer Agent and Custom Agents (GA): Available now in Extend AI, Developer Agent turns plain-language prompts into production-ready agents inside tools like Cursor, Claude Code, or VS Code. It natively understands your tenant’s data model and business logic, extending Workday rather than reinventing it.

• Bring your own framework and model: Build your way. Deploy custom agents built on Google ADK, LangChain (LangGraph), and AWS, powered by models from Anthropic, OpenAI, Google, Meta, Mistral, and NVIDIA Nemotron.

   

Lawful by Default: The Harness Under Every Agent

Base models alone don't solve enterprise compliance. Every agent running on our platform—whether built by Workday, created by your team via Developer Agent, or integrated from a third party—runs on the Workday Agent Harness, GA today.

Out of the box, the Harness delivers five core runtime capabilities:

• Memory: Retains context across turns so agents pick up right where they left off.

• Guardrails: Actively block outside attacks, hidden prompt injections, and off-brand responses.

• Sovereign Routing: Keeps sensitive data within regional boundaries or on Workday-hosted infrastructure.

• Model Choice: Run agents on foundation models from Anthropic, Google, and OpenAI, or open models from NVIDIA, Meta, and Mistral.

• Evals and Traces: Measure accuracy before deployment and give auditors a step-by-step trace of what the agent did.

Coming Soon to The Harness:

• Agent Context Service (ACS): Pre-injects live task context before reasoning begins, cutting token costs and response loops (Early access late 2026, GA early 2027).

• Grounded Reasoning: Validates proposed actions against live tenant policies before execution—stopping unauthorized moves and explaining policy routes rather than breaking compliance (Early access early 2027).

Further, to govern agents at scale, Agent Passport checks them before they ship. Developed with Cisco, Straiker, and OneTrust, Agent Passport issues verifiable Agent Stamps to prove compliance and scan for hidden attacks before production (GA late 2026).

What Lawful Context Enables

Everything enterprise systems know was built for people clicking through screens: institutional memory, org history, audit trails, labor costs, and performance data. Starting today, any agent built anywhere gets to use that live context, safely.

Imagine what that unlocks across your teams:

• Field ops: An agent builds the perfect crew for a job using years of verified skills and performance history—seeing only what it’s permitted to.

• Engineering planning: An agent models the true financial cost of a team's next big bet, right inside Jira.

• Store operations: An agent matches staffing directly to point-of-sale traffic, balancing real-time availability against exact cost-center labor budgets.

   

Go Build

We’ve removed the friction, kept the rules intact, and opened the doors.

1. Head over to developer.workday.com and inspect the specs and MCP catalog.

2. Spin up your personal development tenant.

3. Open Cursor or VS Code, connect Developer Agent, and build your first lawful agent.

Welcome to the open, lawful era of enterprise AI. Let's go build.

Ready to build agents that run on rules you trust? Create your developer account and start building on the Workday Trusted Agent Platform today.