Guardrails for AI agents

Open full demo (opens in a new tab)

Use the left and right arrow keys to move between steps. Enter acts on the highlighted area and Escape closes its tip.

Step 1 of 9: Guardrails for AI agents

Guardrails for AI agents

Agents that call MCP tools act on your data. A guardrail says which tools an agent may use, which fields it may see and how hard it may push, and each agent has its own identity. This tour uses a sample workspace, Acme Demo Corp, with test data only.

Step 1 of 9