Start with a research decision.
Turn a goal into testable hypotheses. Suggestions are starting points for exploration, with their assumptions visible.
Simulation helps surface questions and compare modeled responses. Validate product effects through research with people.
Define the test. See who is in scope.
Keep the proposition, question and context explicit. Interviewed profiles are excluded from every simulation.
A visible path from queue to result.
Live counts describe logical simulated opinions. Fresh, cached, sampled and imputed rows are tracked separately.
Findings, with the evidence attached.
Every answer here is a simulated opinion. Inspect its persona and source IDs before interpreting it.
Ask the next question of a specific cohort.
Retain the parent hypothesis, context and findings. Each follow-up creates its own test and result.
Compare against independent interview evidence.
Benchmark reports separate real interview excerpts, simulated outputs and automated claim judgments. An exploratory product hypothesis has no interview ground truth unless an exact replay is available.
One shared budget. Visible limits.
All paid calls use the project ledger. Mock tests have no model cost; synthetic results remain labeled throughout.