Workday, Inc. (NASDAQ: WDAY) announced at its Workday Rising 2026 conference in Las Vegas that its Trusted Agent Platform now supports headless access, allowing customers to build and run AI agents across HR and finance systems using their own development tools and AI models.

Several components are now generally available, including Agent Harness, Agent-Ready Tools, Workday Data Cloud, Developer Agent, and Custom Agents. Agent Harness runs every agent — whether built by Workday, a customer, or a partner — under the same permissions, approvals, and audit trails already in place within Workday's systems.

Agent-Ready Tools are based on Model Context Protocol (MCP), giving any MCP-compatible agent an approved way to read Workday data and take action. Live Data Query allows agents and teams to query live Workday data in SQL through Workday Data Cloud without custom reports or data copies.

Custom Agents supports frameworks including Google ADK and LangChain, and allows customers to select from AI models such as Anthropic's Claude, Google's Gemini, Meta's Llama, Mistral, OpenAI's GPT, and Nvidia's Nemotron open models. Support for LangChain and Nvidia Nemotron is generally available; Google ADK support is currently accepting design partners.

"Headless can't mean lawless," said Gabe Monroy, chief technology officer at Workday. "The rules for HR and finance already live in Workday, so customers don't have to rebuild them for every agent they create."

Workday said two additional features are planned for Agent Harness: Agent Context Service, expected in early access in late 2026, and Grounded Reasoning, expected in early access in early 2027. Grounded Reasoning would check agent requests against company policies and business processes before an action is taken.

Developer Agent, also now generally available, lets teams describe a desired agent in plain language using tools such as Claude Code, Codex, or Cursor, and generates the agent using Workday's existing security and business processes.