Write what you mean.
Describe the action, the conditions, and when approval is required. Turn supported rules into consistent checks.
Bring in your policiesWrite the rule in plain language. Test how it applies. Bring the right person into the decisions that need approval.
Make policy part of the decision, with consistent checks on work routed through the governed path. Keep the rule, the evidence, and the reason for a hold together.
Bring the policy lifecycle into the work itself. Define what should happen, test the consequence, and keep review connected to the request.
Describe the action, the conditions, and when approval is required. Turn supported rules into consistent checks.
Bring in your policiesReplay a rule against recorded history. Inspect what it would have held before putting it into effect.
Explore simulationGive an authorized reviewer the request, the rule, and the evidence behind the hold.
Follow an approvalEvaluate a draft or saved rule against recorded history. See which requests it would have held or denied, with the reason for each result.
Change the approval threshold to see which requests would need review.
Other permissions and policy checks still apply.
Keep the action waiting while an authorized person reviews it. The approval gate brings the request, the applicable rule, and the supporting context into one decision.
If a required check cannot be completed, the action does not proceed. An approval still has to respect other permissions, policies, and operating limits.
Explore agent governanceKeep each version and the decisions made under it. A new policy changes the work ahead while earlier receipts retain the rule that applied at the time.
Explore receipts and verificationPast decisions keep their policy reference, even after the rule changes.
See the change, the person behind it, and the resulting policy version.
Matching actions are checked against the policy now in force.
Describe the rule in plain language: the action it covers, the conditions that apply, and when it should proceed or require review. Supported rules compile into deterministic checks, with the draft checked for duplicates and conflicts before it is saved. A rule that cannot be interpreted safely does not silently become an active policy.
Yes. Upload a policy document to scan it for standing rules. Review the proposed rules and confirm the ones you want to add. Scanning a document does not activate its contents; confirmed rules pass through compilation and conflict checks and retain their source attribution.
Policy simulation evaluates a compiled draft or saved rule against recorded history in a chosen window of up to 90 days. It shows matching actions the rule would have held or denied and the reasons. It does not execute those actions or activate the draft. Its usefulness depends on the history and context available.
Held actions go through the appropriate approval gate for an authorized reviewer. The gate connects the request, the applicable rule, and the supporting context. The decision and reviewer are recorded. Approval does not remove other required permissions, policy checks, or agent operating limits.
People and agents share the governance path for actions routed through the platform. Applicable policies, permissions, and evidence checks determine the verdict. Agent-specific rules can narrow authority further, but cannot grant authority beyond the organization’s policies.
A policy change creates a new version and a lifecycle record. Earlier versions remain available, and receipts preserve the policy reference behind earlier decisions. An authorized person can also disable, re-enable, or archive a policy through its managed lifecycle.
Required checks fail closed: a rule that cannot be evaluated does not permit the action to proceed. The decision records the problem or missing evidence so it can be addressed. Activity imported after execution is observed evidence and cannot be retroactively held.
Bring a rule, a workflow, and the decisions your team wants to govern.