How it works

01 — Recruit your squad

SQUAT interviews you about the people you already know — how they behave, what they're after, where they lose patience and quit. Most creators and builders know far more about their users than they've ever written down, and the interview is what gets it out. What you describe becomes a persona: a documented perspective with a recorded source.

If you already have research interviews or user profiles, you can load them instead. Detailed profiles come in whole, and SQUAT keeps a record of where each one came from.

Every persona ends with a validation pass: here's what I've got, where am I wrong about them? You know you've got it right when the characterization rings true.

02 — Set a mission

A mission is a goal, not a script. Get this done. No steps, no instructions — because almost nobody reads the instructions.

That's the point. A scripted test can only walk the path you already wrote. A squad member has to figure it out, which is how you find the path nobody planned for.

03 — Run the round

Your squad member goes at the product in character, using the agent and the AI subscription you already use. A moderator runs the room and keeps the session on track.

You see how far your squad members get. You see the dialogue, the thinking out loud, the moment they get stuck, and what they wanted to be different.

04 — Read the ledger

Every round lands on the record — results, findings, the full session, and the model pinned to that run. So when something breaks or seems to go sideways, you can see what changed or fell flat.

Month seven of history is worth more than month one. It compounds, and you can export it at any tier, any time. It also tells your story.

When a client, a peer group, or your leadership asks why you changed direction or made a decision, the record is there to show — the finding, and the session it came from.

Where it runs

SQUAT works inside the agent tools you already work in — Claude and ChatGPT. You work with it and all your content there. There's no separate app to learn.

Interactions with SQUAT run on your own AI subscription. We never meter them and we never bill you for tokens.

What it doesn't do

SQUAT doesn't replace unit tests, integration tests, or harnesses. Playwright and its kin confirm the path you wrote still runs — use them for that. SQUAT is the acceptance layer: what your persona thinks and whether they survive the flow, not whether the flow executes.

Squad members are persona-based AI-simulated testers built from what you know.

Round one won't be perfect. The squad sharpens the way real QA does: run it, correct it, run it again, and be better prepared with the evidence to act.