aolabAgentic Organization LabOFF AIR

AOLab / experimental studio / alpha 01

Watch. Mix.
Replay.

Watch an organization work. Change how it coordinates. Run the same challenge again and inspect every decision.

This is an experiment, not a product demo.In live mode, real language-model agents solve a fictional incident-response challenge. The results are observable but not evidence about human organizations.
LAB 01 / Incident RoomSix fixed cases · reproducible rubricChecking studio status

Incident Room 01

Organizational wind tunnel

A fictional service team must triage six incidents as critical, high or normal. Watch who contributes, who decides and how the final answer is reached.

Not running
Organization topology / hub and spoke
Change the structure to see how information will flow.
AwaitingWorkingFinished

Run measurements

Observed / not simulated
—Priority accuracy
—Critical recall
—Elapsed time
—Model tokens

Rubric: safety or data exposure = critical; widespread outage or billing fault = high; cosmetic or minor messaging issue = normal. Model outputs are not independently verified beyond matching this fixed rubric.

Event stream

Who knew what, when
Studio standing by

Run a live experiment or view the illustrative walkthrough. Recorded events can be replayed and downloaded.

Decisions

Fixed evaluation set

Same six cases across every configuration. The benchmark is intentionally small: useful for inspecting organizational mechanics, not generalizing performance.