Resources · Blog

Field notes on synthetic research.

Opinionated notes on validity, method and product craft from the system that has to preserve disagreement.

Index

Notes8
Lanes4
LatestJun 10, 2026
Product updates

8 notes

Newest first

Jun 10, 2026
Method

The Pollyanna problem with synthetic research

Most AI respondents are trained to be agreeable. Useful synthetic research starts when the room is allowed to object.

Jun 6, 2026
Positioning

Jobs before formats

A focus group, a red-team and a prototype test are not the strategy. They are moves inside a job the buyer actually has.

Jun 3, 2026
Validity

Validity without pretending

Synthetic research should not claim to replace real users. It should make assumptions, risks and next tests visible earlier.

May 30, 2026
Product

Why we built Sonaloop local-first

No new AI workspace to provision, no Sonaloop token meter, and no respondent PII by default. Local-first is a product choice and a research choice.

May 23, 2026
Red-team

Red-team is not negativity

A red-team pass is not the pessimistic person in the room. It is a structured search for the conditions under which the idea fails.

May 16, 2026
Pricing

Pricing research is pressure, not preference

Asking what someone likes to pay is weak. Pricing research gets useful when it exposes budget, trust, urgency and perceived risk.

May 9, 2026
Personas

Personas need memory

A persona without memory is a writing style. Memory gives the panel friction, contradiction and a reason to answer differently next time.

May 2, 2026
Practice

What to do after the council disagrees

A skeptical report is only useful if it changes the next move. Here is how to turn objections into product, messaging and research work.

Threads

Follow one question across the archive.