Answers for teams evaluating Them AI.

Them AI builds synthetic consumer panels from your market signals. You Ask Them to test ads, messaging, pricing, and positioning, verify the read with real humans when the call matters, get Decisions with confidence intervals, then calibrate against real outcomes and external metrics so accuracy compounds.

Them is the people you're selling to - the ones every launch meeting talks about without them in the room. It also spells the four layers of the loop: Twins (consumer twins grounded in your market signals, plus brand twins that ask in your voice - a read in minutes), Humans (your community, Reddit, and interviews of any kind - verification in hours), Evidence (a reconciled ship / iterate / hold call with confidence intervals), and Market (real outcomes and external metrics that calibrate the panel). The journey: test it on twins, hear it from humans, evaluate the evidence, measure it in market.

Them AI is for brand, product, growth, and media teams that need evidence over guesswork before they spend on production, positioning, or paid media.

A panel is a set of synthetic consumers built from your market signals, category context, and segment attributes. It stands in for an audience you can question, not a focus group you have to recruit.

A Decision turns panel reactions into a clear ship, iterate, or hold call with a confidence interval, so you know not just what the panel thinks but how sure to be before you act.

You can test ad concepts, hooks, messaging, pricing, product claims, positioning, audience objections, and segment-level risk before you commit budget.

Sameness - also called ad bleed - is when your variants blur into the same idea: repeated claims, weak differentiation, and reactions that look distinct on paper but land the same with the audience. Because Them AI shows segment-level reactions across variants, it flags sameness before you spend, so you ship ideas that are genuinely different. (Note: this is creative sameness, not in-market ad fatigue, which needs live frequency data.)

Your brand twins - AI agents configured to speak in your brand's voice. They fan questions to your community, moderate Reddit and Slack/Discord threads, and run interviews from a simple link - no scheduling: video, voice, text, or async form, with adaptive follow-ups, and participants can react to the ad or page mid-conversation. Recorded answers are transcribed automatically; it's AI-led, or human-led when you prefer. Everything comes back scored against the synthetic panel. They are always disclosed as AI; the point is scaling honest listening, not impersonation.

Yes. After you Ask Them, you can inspect panel reactions and talk to individual personas to understand why they agreed, objected, hesitated, or preferred another angle.

No. Them AI is pre-spend triage. It produces predictions, not certainties, to help you narrow ideas, find risks, and prepare better tests before higher-cost research or market spend.

Calibration. You log real outcomes after you act, Them AI compares predicted versus actual, and each brand gets an accuracy score that improves over time. That compounding loop is the moat a generic AI chatbot can't match.

Yes. Panels are built from your market signals, so teams can bring research notes, customer attributes, market inputs, and campaign details to shape the panel.

Yes. The loop is composable: stitch steps - twins read, human verification, decision, outcome logging - into reusable workflows that match how your team ships. Templates like ad pre-flight, pricing test, and B2B committee check give you a starting point; edit, reorder, or add steps from there.

Use Decisions as directional evidence: compare options, identify objections, improve creative, prioritize segments, and decide what needs deeper validation. Then log outcomes so calibration sharpens the next call.

A generic chat gives one model's opinion with no grounding and no track record. Them AI builds a persistent panel grounded in your market signals and segments, runs structured multi-perspective reactions, returns a decision with confidence intervals, and then proves itself: every logged outcome scores the panel's accuracy for your brand. Opinions are free; calibrated accuracy is the product.

Building a panel takes minutes. Once a panel exists, an Ask Them run - reactions, objections, segment breakdowns, and a recommendation - comes back in about a minute, fast enough to test while the decision is still open.

Them AI is in early access. We onboard a limited number of design partners and work closely with them - there is no public pricing yet. Request early access and we will reach out with details and a walkthrough.

Simulations run on synthetic personas and anonymized aggregate patterns - raw customer records are not sent to LLMs, and direct identifiers like emails and phone numbers are excluded before model calls. Storing real prospect details is a separate, explicit opt-in per workspace, and you can delete any record at any time.