Your AIcan act. Who gets hurt?

A bespoke tabletop exercise for the AI you run. We find the weak spots, with care.

Built around your environmentPlayed by your teamHalf or full day

// 02 / why pink02/10

Security teams come in colors.

RedFinds the break.
BlueHolds the line.
PurpleGets them talking.
PinkAsks who gets hurt.

Most tests ask how it breaks. Pink also asks who it hurts.

Red, blue and purple are established practice. Pink is new, and we say so on purpose.

// 03 / the table03/10

Three seats at the table.

It is a tabletop exercise. Your own people take the seats and play a scenario out, move by move, on a system you actually run.

AdversaryPlays the attack.
OperatorRuns the system and lives with it.
Person harmedSpeaks for whoever is on the other end.

Most exercises fill two seats. The third one is pink.

// 04 / the scenario04/10

Built around your world.

Nothing here comes off a shelf. We learn how your system works, then write the exercise around a situation that could really happen to you.

We learnYour system, your people, what it can touch.
We writeA scenario that makes sense for you.
We buildThe cards your table will play it with.

Every table plays a different game. Yours is made for you.

// 05 / the board05/10

Rank by harm.

Every weak spot the table finds goes on the board. Slide one toward the people it reaches and watch the order change.

Illustrative examples. Your session writes its own.

// 06 / the session06/10

One session. Four moves.

MapYour system and everything it can touch.
PlayThe scenario, live, with your people in the seats.
RankEvery weak spot, ordered by who gets hurt.
KeepA threat model your team built itself.

Runs as a half day or a full day.

// 07 / the value07/10

What you leave with.

A threat modelOf the system you actually run.
A ranked listWeak spots, ordered by who gets hurt.
A shared pictureEveryone saw the same thing happen.
The practiceYour team has now done this once.

The value is in the playing. Your team does the threat modeling.

// 08 / the price08/10

What it costs.

One price per session, your whole team in the room. Each one is designed for your environment.

Half-day workshop$6,000
Full-day workshop$12,000
Larger teams and repeat sessionsAsk

This is adversarial testing of how your AI behaves, and practice for the people around it. It is not a compliance audit, an infrastructure pentest or a guarantee.

// 09 / the manner09/10

Weak spots, with care.

Finding what could go wrong is the job. How it is done matters just as much.

No gotchasNobody is being tested. The system is.
No blameA weak spot is a finding, not a fault.
Your paceThe table decides how far to push.

The point is to see clearly. Then to look after the people on the other end.

// 10 / your move10/10

Tell us what you run.

A few lines is enough. We will come back with a scenario shaped around your environment.