Workday Opens HR and Finance Data to External AI Agents with Headless Platform

Workday's Headless Platform Launch

Workday announced at its Workday Rising conference that its Trusted Agent Platform is now "headless." This development allows external AI agents to directly access and interact with Workday's HR and finance data. The platform's Model Context Protocol (MCP) server, SQL access to live Workday data, and a shared runtime for agents are now generally available.

Expanded AI Agent Development

The headless platform enables developers to build their own HR and finance agents using various AI models from providers such as Anthropic, Google, Meta, Mistral, Nvidia, and OpenAI. These agents can be developed with tools like Claude Code and Cursor, expanding the ecosystem for custom solutions within Workday.

Workday's Agent-Ready Tools, including the MCP server as part of Workday Extend AI, serve as the primary entry point for AI developers. The server currently offers 275 read tools and 54 tools for writing or deleting data.

Ensuring Data Security and Control

A core aspect of this update is the enforcement of Workday's existing security protocols. All agents, regardless of who builds them, operate under the same Workday permissions, approvals, and audit trails as human users. This prevents agents from bypassing established access controls, a concern Workday highlights regarding other methods of connecting agents to HR and finance data.

Live Data Query, a component of Workday Data Cloud, allows business intelligence tools and agents to execute SQL queries directly against Workday's data. Each query is designed to return only the data the requesting user is authorized to view, maintaining data integrity and security.

Shared Runtime and Agent Creation

The Agent Harness provides a shared runtime environment for both Workday's internal agents and those developed by customers and partners. This runtime passes the user's identity, organizational position, and access rights to each agent, limiting its actions and selecting the appropriate AI model based on cost and performance.

Workday also introduced Developer Agent, which generates Workday agents and applications from natural-language prompts in various coding environments. Additionally, Custom Agents allows teams to deploy their self-built agents within the Workday ecosystem.