Scale AI execution without giving up control

Watch authority become a visible part of the work.

Agents keep working. Crenovent makes the boundary explicit: what can proceed, what must wait for review, what is blocked, and the evidence attached to every decision.

Authority Sandbox Not “human in the loop” as a slogan. This experience lets the visitor put live agent actions through the operating boundary and see exactly what happens.

Revenue Agents & Automation· operating within configured authority

Auto demo · runningCommercial situation · Synthetic demo

Just in time

Policy exception needs review

An agent action reached a boundary that requires an authorized decision before anything can proceed.

Tell Me · authority context

This action is attached to the same commercial situation and authority state currently visible in the sandbox.

Authority Airlock

One action at a time. No invisible autonomy.

The same agent can read, prepare, recommend or execute only as far as the active policy allows. Change the action or authority profile and the boundary recalculates.

Allowed

Proceed within authorityPermitted action · evidence retained

Approval required

Hold for authorized reviewNothing executes before decision

Blocked

Do not executePolicy reason preserved in audit

Meeting intelligenceEvaluating

Create a follow-up action from the meeting

Owner, due date and meeting evidence are available. No sensitive customer communication is sent.

Authority profilePer capability · policy still applies

Current profile: proposals are allowed; consequential execution remains review-gated or blocked according to policy.

Production proof drops into the same composition.

S14 product proof is still pending. These apertures preserve the prototype architecture while giving the developer exact places for approved product captures when they exist.

P14-RA-01

Revenue Agents & Automation

Agent operating status, capability state and bounded execution controls.

Required provenance: approved actual product capture or explicitly labelled source-derived staging.

P14-AC-01

Action Center review

Scenario-matched approval-required decision with evidence, actor and outcome state.

Do not reuse renewal-specific S6 proof as generic agent-governance evidence.

P14-GV-01

Authority / policy control

Approved control showing the configured execution ceiling, blocked reason or operating guardrail.

If no live screen exists, retain this named aperture rather than fabricate product UI.