POPULATION INTELLIGENCE · MARKET VALIDATION
Validate demand before you build. Ask a census-grounded crowd, not an idea scorer.
An AI agent runs the read mid-task over the CrowdOS MCP server, on every plan, Free included. Ask a panel drawn from census data that disagrees with itself. What comes back is reasoned reactions, with verbatims — not one model's opinion.
WHO YOU ARE ASKING
A census-grounded crowd, not an idea scorer.
A panel drawn from census data disagrees with itself — not one model's opinion. An LLM persona is a costume for one prompt. Ours is a database row. CrowdOS agents are persistent database rows — census-grounded people with a fixed identity, an opinion history, and a standing news habit — not LLM personas spun up per prompt.
Every respondent is drawn from demographic distributions calibrated against published benchmarks: US ACS 2023, UK ONS 2021, AU ABS 2021. Large language models are trained to be helpful. The anti-sycophancy layer counteracts that default in every study. Respondents are permitted — and instructed — to be critical, contradictory, and specific.
THE WORKFLOW
Four reads, in order, then a later check on a fresh sample. Same persistent population. Starter templates across 24 markets, or a custom segment by country, age, gender, income, education, and political leaning.
01 · CONCEPT
Independent Vote.
Each respondent answers independently — no inter-agent influence. Output aggregates to a sentiment distribution, demographic crosstabs, and verbatim excerpts. This is the first read on a concept.02 · OPTIONS
Comparison and MaxDiff.
Respondents rank from their persona context, values, and category knowledge. Ranked-choice aggregation produces a ranked verdict with attributed reasoning per option. Headlines, taglines, and value propositions, with verbatim reasoning on which language resonates with which segment. MaxDiff ranks messages, claims, reasons or criteria by best-worst scaling: each respondent sees a few small sets and names the most and least, and you get an order with confidence intervals.03 · PRICE
Van Westendorp and Gabor-Granger.
Van Westendorp maps the acceptable price range; Gabor-Granger finds the revenue-maximising point on a ladder. Same synthetic panel, classic instruments. Van Westendorp maps the range a price has to sit inside to read as neither a rip-off nor suspiciously cheap; Gabor-Granger walks a ladder to find the revenue-maximising point. The study returns price points and a revenue curve.04 · MOCKUP
Screening Room.
Upload a creative, product mockup, policy draft, pitch deck, or document. A representative synthetic panel reacts and explains why, with cohort breakdowns and expert lenses.LATER
Re-field on a fresh sample.
Re-field the same instrument as the world moves. Each study draws a fresh, independent sample — drift comes from real-world shift, not from the same panel changing its mind. Each study draws an independent, freshly-sampled panel from the crowd. The crowd evolves only from exogenous signal: news consumption, life events, peer interaction. It does not evolve from being asked.
WHAT YOU GET BACK
Reasoned reactions, with verbatims.
Reasoned reactions from a census-grounded panel, with verbatims. A sentiment distribution, demographic crosstabs, and verbatim excerpts. The study reports what the panel concludes. Purchases and signups are a separate record.
WHAT THE STUDY REPORTS
The panel's conclusion, on the record.
CrowdOS is the crowd you ask before you build. Message-market fit and high-frequency creative evaluation come back as reasoned reactions from a census-grounded panel, with verbatims, without the fielding-time and recruitment-cost constraints that make traditional qualitative research impractical in product-launch contexts. Purchases and signups are recorded elsewhere.
THE OPINION BENCHMARK
The published opinion benchmark.
On the Pew opinion benchmark, the crowd's opinion distribution matches published polling at 93.3% parity, with a mean absolute error of 3.36 percentage points and a Pearson r of 0.984. That figure measures opinion. Purchases and signups are a separate record.
WHERE THIS FITS
The same crowd, asked a different question.
Pulse is a vote, a debate, or a comparison. Screening Room pressure-tests a creative, a concept, or a deck before you ship. Pre-launch testing is the step before a live A/B test. The MCP server lets an agent run the same studies.
FAQ
What does market validation look like here?
Start with an Independent Vote: each respondent answers independently, and the output is a sentiment distribution, demographic crosstabs, and verbatim excerpts. Rank options with Comparison or MaxDiff. Read a price with Van Westendorp and Gabor-Granger. Put a creative, product mockup, policy draft, pitch deck, or document in front of Screening Room.
Is CrowdOS a substitute for customer research?
CrowdOS is the panel you ask before you build. Message-market fit and high-frequency creative evaluation come back as reasoned reactions from a census-grounded panel, with verbatims, without the fielding-time and recruitment-cost constraints of a traditional qualitative study.
What kind of reactions come back?
Reasoned reactions from a census-grounded panel, with verbatims. The study reports the panel's conclusion. Purchases and signups are a separate record.
Which study fits which step?
Pulse Independent Vote reads a concept. Pulse Comparison ranks headlines, taglines, and value propositions. MaxDiff orders messages, claims, reasons or criteria. Van Westendorp and Gabor-Granger read a price. Screening Room reads a creative, product mockup, policy draft, pitch deck, or document.
Can the same question be checked again later?
Re-field the same instrument as the world moves. Each study draws a fresh, independent sample — drift comes from real-world shift, not from the same panel changing its mind.
Ask the crowd before you build.
Reasoned reactions from a census-grounded panel, with verbatims.