Data Scope
Only the tenant, learner, case and fields required for the task.
Explore →Agent Governance
Give agents permissions smaller than the people who supervise them.
The platform formalises agent permissions, risk tiers, tool allowlists, data scope, action budgets and escalation rules.
Tool Allowlist
Agents may only call explicitly approved adapters.
Explore →Financial Limits
No autonomous discretionary refunds or unapproved price changes.
Explore →Execution Limits
Bound retries, run duration and queue depth.
Explore →Reversibility
Prefer reversible steps until human approval is granted.
Explore →Run Ledger
Every significant step is attributable and reviewable.
Explore →Kill Switch
Global and tenant-level disable controls.
Explore →Evaluation
Test workflows before enabling production automation.
Explore →