Agent Harness

Agent Harness is the same infrastructure within which all agents on Workday run. It provides context, memory, guardrails, and reasoning for that work. You write the agent logic, and we maintain its infrastructure, so you can focus on building the agent rather than running its infrastructure.

Agent Harness runs agents, so you can write the agent logic without running its infrastructure. Each agent runs in its own fully isolated environment to help protect your data. You can focus on what the agent needs to do inside Workday while we run it.

Agent Harness enables agents to work from your live organization, approval chains, and policies as configured. Its Context Management capability gives agents information at the right time to help them reach an answer. Agents can work with how your organization is configured today rather than relying on an older copy of those details.

Agent Harness includes Memory, which lets agents retain information across conversations. For agents doing work inside Workday, information from one conversation can carry into another. Memory works alongside the context, guardrails, and reasoning Agent Harness provides, rather than confining an agent’s work to a single exchange.

Agent Harness uses guardrails and scoped execution to keep agents from accessing data the person using them cannot access. The guardrails also protect agents against attacks. Each agent runs in its own fully isolated environment to help protect your data while it does work inside Workday.