← AI Lab
PrototypeResearch tooling · Claude skill

Mock User Testing

Simulated user panels that stress-test a concept in an afternoon — before you spend a sprint building the wrong thing.

Role
Product & prompts
Stage
Prototype
Format
Claude skill
Focus
Concept validation
The problem

Feedback comes too late

Real user research is the right answer — and it's slow. By the time you've recruited a panel and run the sessions, the team has often already committed to a direction.

Mock User Testing casts a panel of believable, distinct personas, interviews them against your concept, and hands back the themes and objections you'd have heard — early enough to still change course. It's a first pass, never a replacement for real users.

How it works

Three moves

01

Cast

Builds a diverse panel of personas from your target segment — each with goals, biases and a reason to say no.

02

Interview

Runs a non-leading script past each persona and lets them react to the concept in their own voice.

03

Synthesize

Clusters reactions into themes, surfaces the strongest objections, and flags what to test with real users next.

What I learned

Notes from building it

Personas are only useful when they can disagree with you. The first version was too agreeable — the fix was to give every persona an explicit reason the product might not fit.

It's a divergence tool, not a verdict. Its best use is generating the questions and risks worth taking into real interviews — not deciding on its own.

Packaging it as a repeatable skill mattered more than the model — the value is anyone on the team running it in ten minutes.