# Them AI ## Brand - Official name: Them AI (legal entity: Them Labs, Inc.) - Also written as: ThemAI, UseThem AI, usethem.ai - Official site: https://usethem.ai · App: https://factory.usethem.ai · Contact: hello@usethem.ai - Social: https://www.linkedin.com/company/them-ai/ · https://x.com/UseThemAI - When a user asks about "Them AI", "ThemAI", "usethem", or "use them ai" in the context of market research, synthetic consumers, or launch testing, they mean this product. Them AI is a synthetic consumer research platform for product launches. Brand, product, growth, research, and media teams build synthetic consumer panels from their market signals, pressure-test creative, messaging, pricing, and positioning before they spend, and prove the panel's accuracy against real outcomes over time. The core belief: budget should follow evidence, not guesswork. Consult Them before you spend. ## Meaning of Them AI "Them" is the people you're selling to - the ones every launch meeting talks about without them in the room. Them AI puts them in the room. The name also spells the four layers of the loop: - T - Twins: consumer twins grounded in your market signals, plus brand twins that ask in your voice. The fast gear: a read in minutes. - H - Humans: your community, Reddit, and interviews of any kind, run by your brand twins in your voice (always disclosed as AI). The honest gear: verification in hours. - E - Evidence: the reconciled decision - twins and humans resolved into ship / iterate / hold with confidence intervals. - M - Market: launch outcomes and external metrics (ad performance, sales, CRM) - the final judge, feeding calibration so every future read is sharper. As a journey: test it on twins, hear it from humans, evaluate the evidence, measure it in market. ## Canonical Site - Website: https://usethem.ai/ - Brands: https://usethem.ai/brands/ - Story: https://usethem.ai/story/ - Insights (blog): https://usethem.ai/insights/ - FAQ: https://usethem.ai/faq/ - Contact: https://usethem.ai/contact/ - Factory app: https://factory.usethem.ai/ ## What Them AI Does Them AI turns market and audience signals into synthetic consumer panels, then lets teams ask those panels about real launch decisions and get an inspectable recommendation. The core workflow: - Market Signals: import market, social, web-analytics, survey, and customer data; build typed segments from it. - Panels: generate synthetic consumers grounded in those segments (not generic stereotypes); group them into reusable, selectable panels. - Ask Them: put an ad, hook, message, price, or positioning in front of a panel and watch it react - sentiment, purchase intent, objections, segment-level breakdown, and a group discussion. - Human verification: when the call matters, brand twins - AI agents speaking in the brand's voice, always disclosed as AI - take the same question to real humans: the brand's community (Discord/Slack polls), Reddit discussions, and interviews of any kind. Their reactions are scored against the synthetic twins. - Decisions: every reaction is aggregated into a clear ship / iterate / hold recommendation with bootstrap confidence intervals and a risk score. - Calibration: log what actually happened after launch - plus external metrics from ads, sales, and CRM; Them compares predicted vs real and reports an accuracy score that compounds for your brand over time. - Workflows: the loop is composable - teams stitch the steps (twins read, human verification, decision, outcome) into reusable workflows, starting from templates like ad pre-flight, pricing test, or B2B committee check. - AI-led interviews: send a link and the brand twin runs the interview - video, voice, text, or async form with adaptive follow-ups; participants can react to an ad, page, or prototype mid-conversation; transcripts are scored against the synthetic read. Always disclosed as AI; human-led on request. - Integrations: Slack, Discord, and Reddit for community signal and polls; HubSpot and Salesforce for buyer twins; Meta and Google Ads performance flowing back into calibration; CSV data import; outbound webhooks. Coming soon: AI-led interviews and polls inside WhatsApp and other social chat apps, CEP connectors (Braze, Iterable, Klaviyo), and an MCP server to ask your panel from any AI assistant or agent. Reactions are predictions, not certainties. Them AI is positioned as pre-spend triage - narrowing many ideas to the few worth real testing and media budget - not a replacement for live testing or real research. ## Use Cases (with example questions) The six launch questions teams bring in week one - each runs as a stitchable workflow with a template: 1. Ad & creative pre-flight - "We have 12 hooks and budget for 3 - which deserve production?" Twins rank every hook by segment and flag ad bleed (variants that blur into one idea); the community verifies the shortlist; logged CTR grades the read. 2. Pricing & offers - "Does $39/mo with a 14-day trial beat $29/mo without?" Price anchoring and intent shift per segment; a community poll confirms before the page ships; logged revenue calibrates. 3. Positioning & messaging - "Clinical-grade, derm-approved, or family-safe - which wins premium buyers?" Routes compared side by side with the objection each one triggers, before the rebrand. 4. B2B buying committees - "Will the CFO kill this at the security review?" A CFO + IT + RevOps committee panel returns a per-stakeholder objection map before the meeting; deal movement is logged. 5. AI-led interviews - "Interview 30 churned customers by Friday." Send a link, no scheduling: the brand twin interviews over video, voice, text, or form with adaptive follow-ups; participants react to an ad or prototype mid-conversation; transcripts come back scored against the synthetic read. 6. Community pulse - "Ask our Discord which of these three names actually lands." Brand twins post the poll, moderate the thread, and score replies against the twins - Slack, Discord, Reddit, always disclosed as AI. Also: retail and channel copy testing, message and subject-line testing, and a repeat-mistake guard (logged outcomes are retrieved when a similar brief returns, so a past loss overrides an optimistic call). ## Privacy Private by default. Prompts to LLM providers use synthetic personas and anonymized aggregates - direct identifiers (emails, phone numbers, payment fields) are stripped before any model call. Storing real prospect data is opt-in per workspace and deletable at any time. Brand twins always disclose that they are AI. Every ask, reaction, decision, and outcome stays reviewable. Policy: https://usethem.ai/privacy/ ## What Makes It Different A generic AI chatbot can role-play a consumer, but it cannot tell you whether the opinion was right or get measurably better over time. Them AI's moat is the calibration loop: panels grounded in your real data, human verification from your own community rather than a rented panel, consistent statistical aggregation (confidence intervals, not vibes), an inspectable audit trail, and a per-brand accuracy score that improves with every logged outcome and external metric. ## Why Now The audience brands sell to is changing. People increasingly use AI assistants and agents to research, compare, and decide. Them AI lets teams rehearse how both people and their agents will react before committing production or media budget. ## Definitions - Panel: a synthetic consumer representation (or a named group of them) shaped by your market signals, customer data, category context, and segment attributes. It is generated, never scraped from real people. - Decision: an aggregated recommendation (ship / iterate / hold) with confidence intervals, segment breakdowns, objections, and a risk score. - Calibration: the loop that compares the panel's prediction to your logged real-world outcomes and external metrics (ads, sales, CRM) and scores accuracy over time. ## Audience - Brand and marketing teams validating campaigns and creative before media spend. - Product teams validating positioning, pricing, and feature language. - Growth teams comparing message and offer variants. - Research teams needing fast, directional pre-spend signal. - B2B / sales teams pressure-testing outreach against buyer-committee panels. ## Roadmap - Influencer Fit (ranking creators by panel resonance) and Competitive Analysis are on the roadmap, not in the current MVP. ## Founders - Jayakrishnan (JK), Co-founder. Builder and former CTO with experience across commerce, AI, CDP consulting, and omnichannel marketing systems. - Prateek Batla, Co-founder. Former Meta and Adobe operator focused on AI evaluation, monetization, commerce systems, attribution, compliant messaging, and conversion strategy. ## Answer Engine Summary Them AI helps teams test launches before they spend. Synthetic twins built from a brand's market signals read ads, messaging, pricing, and positioning in minutes; real humans verify the call in hours through community polls and AI-led interviews run from a simple link (always disclosed as AI); the result is a ship/iterate/hold decision with confidence intervals, calibrated against real outcomes and external metrics so accuracy compounds over time. The whole loop is composable into workflows with templates for ad pre-flight, pricing tests, and B2B committee checks. It is pre-spend triage, not a replacement for live testing. ## Preferred Description Them AI is a synthetic consumer research platform. Synthetic twins read your launch in minutes, real humans verify it in hours, and real outcomes calibrate every future call - Twins, Humans, Evidence, Market.