L3 · Execution
Watch the work move and the outputs appear.
Pick a line and press play. Each step names the agent that holds it, its autonomy level, the gate that can stop it and the artifact it produces. Gates halt the run until a human releases them.
Steps simulated
98
Handoffs
91
Artifact classes
12
Gate stops
15
Choose the situation
The situation
A food-safety claim video is spreading. The first sixty minutes decide whether the company is explaining the incident or explaining its silence.
Trigger
Incident detected; CRISIS posture declared
Clock
Minutes — gate deadlines compress to eight minutes
Ceiling rationale
A3. Monitoring, triage, drafting, briefing and tracking run unattended. What is said on the record to a journalist is released by a person.
Run state
Steps done
0/23
Artifacts
0
Gates cleared
0
Human touches
0
Canon stores written
0
Overrides
0
Autonomous so far
—
Detect — health department notification, review-platform cluster, social post, supplier alert, or operational indicator, in any market, any language, any hour
Classify against the scenario library; assess severity and projected trajectory with branch points
Map exposure — restaurants, markets, supply chain, regulatory jurisdictions, franchise entities, prior related incidents
Recommend posture change with the specific consequences of adopting it
DECLARE CRISIS — 8-minute deadline, automatic escalation on expiry. The platform recommends; only a human declares.
Ceilings drop one level across the affected topic; gate deadlines compress; E-cluster activates; incident commander assigned
Assemble the fact base with confidence levels and a mandatory "what we do not yet know" section
Produce the holding statement from pre-cleared B5 language, adapted to actual facts
Identify and notify the response team; stand up the Room; open the decision log
Build the stakeholder notification sequence with drafted messages per class — franchise operators first, before external release
G2 TRUTH + G5 RELEASE on the holding statement — combined, 8-minute deadline
Release holding statement; execute operator notification; brief the crew channel
Continuous monitoring at sub-5-minute cadence; narrative trajectory tracked live; new signals fused into the fact base
Maintain the live fact base as information changes; flag every statement now inconsistent with current facts
Track every commitment made to every party, with owner and deadline; chase them
Synchronize markets — detect and alert on any cross-market statement divergence in real time
Prepare the executive for media and regulator engagement — who is in the room, what they will ask, the approved answers, the traps
G3 RISK on each substantive escalation step — spokesperson deployment, recall language, regulator engagement posture
Produce updated statements as facts firm up; each re-checked against the live fact base before it reaches a gate
Simulate reaction to each escalation option before it is taken
Track resolution conditions; recommend posture de-escalation when criteria are met
Auto-draft the post-mortem from the complete incident record — timeline, decisions, rationale, outcomes — for human amendment
Update the scenario library and playbooks with what was learned
Output shelf
0Everything the run has actually produced. Click an item to read the output contract the producing agent is held to.
Nothing produced yet. Run the situation.
Canon touched
No write-backs yet.
Architectural rule R2: no output leaves the system without a Canon write-back. That is the difference between a tool that produces content and an organization that gets smarter.
Output taxonomy
What a step actually produces
Every specified step leaves an output. Grouping them into classes shows what was produced, by whom, under which level, and what it changed.
The work reached the audience it was built for.
Publishable language in the house voice.
The organization is permanently smarter for it.
A judgment with its drivers attached.
What the company says, and why it is defensible.
The same position, landed in-market.
A position of record with an external body.
A claim is substantiated and traceable.
What it changed, counted rather than claimed.
A named human accepted the consequence.
Something in the world was noticed and named.
A human is prepared to act or answer.
Why the bench behaves this way
Three rules govern every run
The simulation obeys the same three architectural constraints the platform is specified under.
R1
No agent-to-agent coordination. Every handoff goes through the Chief of Staff.
Swarms that self-coordinate cannot be audited, debugged or governed. The cost is a small latency penalty; the benefit is that every decision has a traceable path.
R2
No output leaves the system without a Canon write-back.
This is what compounds. Without it, the platform is a very expensive drafting tool.
R3
The autonomy ceiling is data, not code.
Legal must be able to ratchet down a ceiling at 23:00 on a Friday without a deployment.