Research · notes on synthetic research

What holds up in synthetic research,and what does not.

Notes on synthetic research from the people building a persona simulation platform: what the terms mean, what the evidence says, how to run it so the output holds up, and what to ask a vendor. Published when there is something true to say, with the sources attached.

TopicsWhat synthetic research isDoes it hold upGrounded and auditablePlaybooksBuyer’s guide
R-10

What is persona simulation? The idea, its research lineage, and where it stops

Conditioning a language model on a detailed persona and letting it answer as that person. Where the idea came from, what the research calls it, where it stops.

Definitions02 Oct 2026 · 8 min
R-11

The 85 to 95% accuracy claim, unpacked: seven numbers, seven different measurements

Every vendor quotes a number in the same band. Each one measures something different on a different task. Where each figure comes from, and what it can carry.

Evidence02 Oct 2026 · 9 min
R-01

What are synthetic users? A plain definition, and what the term hides

AI-generated personas that stand in for research participants. What the term covers, the four kinds of system behind it, and where the definition stops.

Definitions23 Sept 2026 · 8 min
R-02

Are synthetic respondents accurate? What the evidence says, both ways

Vendors quote 85 to 95%. The peer-reviewed record is more specific: aggregates often match, variation does not, and it rests on what the persona was built from.

Evidence23 Sept 2026 · 9 min
R-03

Grounded vs prompted AI personas: the difference that decides whether the output is evidence

Prompted personas answer from general knowledge. Grounded ones answer from your material, cite it and decline beyond it. Why that decides what a reply is worth.

Method23 Sept 2026 · 8 min
R-04

Why AI personas agree with everything, and how to build ones that push back

Sycophancy is how assistant models are trained, not one product’s bug. What the research says, why it flattens findings, and three mechanisms that counter it.

Evidence23 Sept 2026 · 7 min
R-05

How to pressure-test pricing with AI personas before you spend on fieldwork

A playbook for a pricing question: which personas to build, what to ask, what to read in the trace, and how to turn a run into a report someone can defend.

Playbook23 Sept 2026 · 8 min
R-06

Synthetic users vs real users: what each is for, honestly

Synthetic users are fast and cheap; real users are the ground truth. A comparison by question type, with the evidence on where simulated answers track people.

Comparison23 Sept 2026 · 7 min
R-07

What is a synthetic focus group, and when does it beat a room of eight?

Two to eight AI personas, a moderator or none, replies that react to each other. What it is, what it is good for, and the failure modes to design around.

Definitions23 Sept 2026 · 7 min
R-08

How to validate synthetic research: a concordance method anyone can run

Do not take a vendor’s number. Re-run a study whose human answer you hold, compare distributions not averages, and read the trace where they diverge.

Method23 Sept 2026 · 8 min
R-09

Fifteen questions to ask a synthetic research vendor, with the answers to listen for

The questions that separate a research instrument from a chat with a system prompt: source material, grounding, traces, refusal, validation and data handling.

Buyer’s guide23 Sept 2026 · 8 min

Private beta · by application

Bring one real research question.

Access opens by application. Tell us the question you are trying to answer and who has to be convinced by the result.

Apply for accessRead the method

A rehearsal, not a replacement, for talking to real customers. Stated in the product and in the terms. Every exported report ends with the line that personas are AI simulations, not real customers.