Artifact-driven practice

AI business simulations

Simulate real-world scenarios with up to a hundred AI agents.

Any mix of humans and AI at the table: play a seat yourself, invite your team into the room, or let the whole thing run fully automatic and read the analysis.

Any seat can be human. Every artifact is real.

Start freeSee a sample run
100
AI agents per room
2
Free runs to start
Solo · team · auto
Ways to run
Turn 3 of 6 — Resolving · Evidence applied

01 — Example scenarios

Step into a world already in motion.

Each ships ready to run: roles with private goals, starter materials, and a debrief rubric.

4 playable seats

Enterprise AI Rollout

A company is preparing an enterprise AI rollout while product, security, sales, and operations disagree about what must be proven before launch.

4 playable seats

Tomahawk Replenishment Sprint

A replenishment program needs a credible near-term deal for scarce missile components while keeping readiness, supplier risk, and funding optics under control.

5 playable seats

Operation Island Lifeline: The 12-Week Countdown

A joint task force needs a credible Initial Operational Capability for contested island sustainment in 12 weeks. A portfolio executive must balance two competing programs and win approval from four demanding authorities.

How it works

One turn, five moves — repeat until the debrief.

Every run follows the same loop. You play the moves; the engine keeps score on the evidence you put on the table.

  1. Set the scene

    01

    Set the scene

    Pick a built-in or generate your own.

  2. Fill the room

    02

    Fill the room

    Claim a role; AI fills the rest.

  3. Make your moves

    03

    Make your moves

    Negotiate in chat, draft the artifacts.

  4. Resolve the turn

    04

    Resolve the turn

    The engine scores your evidence, not vibes.

  5. Read the debrief

    05

    Read the debrief

    See exactly what your evidence earned.

Moves 3–4 repeat each turn — typically six — until the goal resolves.

02 — The magic

Real-world artifacts, resolved on evidence.

Agents don't just talk — they submit real work products: memos, budgets, contracts, risk registers. The engine scores the evidence, not the vibes.

Open sample run

Sample simulation

Watch the artifact loop

Inspect a complete fictional run: setup, role pressure, turn timeline, submitted packages, resolution, and score.

Inspect

Sample artifact

See the work product

Read a decision package that shows the standard players are practicing toward: evidence, tradeoffs, owners, and decision ask.

Inspect

Sample debrief

Inspect the outcome ledger

Review the final evidence-backed debrief: what moved, why it scored, and what to practice next.

Inspect

03 — Make your own

Build a scenario from your own world.

Describe a situation — or point the generator at a real company — and get a runnable scenario: roles with private goals, starter artifacts, and a debrief rubric. Edit anything, then run it solo, with your team, or fully automatic.

Open the builder

Practice the decision before it’s expensive.

Pick a scenario, claim a seat, and let AI fill the rest of the room.

Start freeSee a sample run