Your users aren't in the room.
Interview them anyway.

PersonaLab builds a pool of personas grounded in real U.S. demographics, runs a structured study with the ones that fit what you're building — up to ten at once — and hands you a decision brief: how they handle it today, the quotes behind that, and the next product move.

no card needed · building your pool is free · a study costs 1 credit per persona · how it works ↓

How it works

Three steps. About five minutes.

No research background needed. You describe, the personas talk, you get something you can act on.

1
Build a pool

Your product + your target

Say what you're building and who you want to learn from. We narrow to that demographic and rank every persona by how many of your requirements they meet — must-haves first — so you can see who's worth asking. Up to 100 in a pool.

2
Interview

Ask up to ten at once

Pick what you want to learn. Explore the problem runs a fixed six-question guide — how each persona handles the activity today, where it fits in their life, what shapes it, what they use, how regularly, what they'd change — and your product is never mentioned. Concept, pricing and open studies send your own questions to everyone at once. Up to ten questions per study.

3
Report

A decision brief, not a transcript

One proposed next move — what to build first, for whom, and the in-product signal that confirms or reverses it. Key findings with the quotes behind them, the full interview per persona, and a read-only link you can share. Concept and pricing studies still count who said they'd use it (19 of 30, never a projection).

Why you can trust it

Synthetic, but not made up.

The fair question about AI personas is “why would I believe them?” Here's the honest answer.

Grounded, not invented
Personas come from real U.S. demographic distributions (NVIDIA Nemotron-Personas) — age, occupation, location, education — not improvised by a chatbot on the spot.
Cited, not vibes
Every finding shows the quotes behind it — which persona, which question. A quote lifted from a stored profile opens the exact source passage; one the persona reasoned out from its profile is labeled inferred from profile. Decide for yourself. No “trust me.”
Directional, honestly
We never dress this up as statistics. It's a fast, structured way to pick a first move before you spend weeks building the wrong thing.

In your editor

Or ask them without leaving Claude.

Already working in Claude Code, Cursor, or Codex? Connect PersonaLab over MCP and your personas become people you can talk to mid-build. Casual interviews are free; a full study costs the same 1 credit per persona it does here.

claude mcp add --transport http --scope user personalab https://personalab.aimgonna.com/api/mcp

See what your users would say.

PersonaLab — interview your users before you build