See how customers react before you ship
It's AI-powered user testing with synthetic testers trained on your customer data, running inside AI coding/design tools (Claude Code, Cursor, v0, Lovable) and ticketing tools (Jira, Linear) so customer validation happens automatically in your workflow. Delivers recorded think-aloud sessions, AI-moderated tests, and insights in minutes.
Custom personas trained on your customer data (call recordings, surveys, support tickets, CRM) that use your product, share their screen, and think aloud — recorded like a real session. Built on a base model trained on 100,000+ hours of real usability sessions.
Test with your own customers, testers recruited from TheySaid's panel, or third-party panels — screened, scheduled, and paid for you, with results in hours.
An expert critique of the experience against your spec and best practices before anyone else sees it.
Runs inside Claude Design, Claude Code, Cursor, Replit, v0, Bolt, and Lovable via TheySaid MCP, and triggers automatically when a Jira ticket reaches 'Design Done', attaching evidence to the ticket.
Run user tests, interviews, AI chat, polls, or surveys from one interface, with conditional logic, templates, and localization in 70+ languages.
Automatic insights within minutes: usability issues, behavior patterns, task success rates, transcribed/timestamped voice feedback, highlight clips, and cross-project insights via 'Ask AI anything'.
Screeners keep the wrong people out of human tests; Response Quality Monitor filters low-effort answers before analysis.
One person connects TheySaid to your existing tools (Claude, Cursor, Jira, etc.) in under 2 minutes; no new logins for the team.
Pick your own customers, recruited human testers, synthetic audiences trained on your data, or the Expert AI Reviewer — and compare roles like buyer, end user, and exec in one test.
Users or synthetic testers complete guided tasks with screen and voice recording. AI analyzes the results, surfaces issues, and returns a fix prompt you can paste into Claude or your editor.
Free plan at $0/month includes 5 AI user tests, 50 AI surveys, 1 workspace and 1 active test (panel recruitment costs extra). Enterprise is custom-priced, billed annually, with unlimited tests, workspaces, SSO/SAML, CRM integrations, and dedicated support. Synthetic tests start at $1 each; extra fees apply for hard-to-recruit testers.
$0/month
5 AI user tests/month, 50 AI surveys/month, 1 workspace & 1 active test, AI project creator, AI test moderator, AI analytics & reporting, shareable link. Panel recruiting at extra fee; AI responders coming soon.
Custom (billed annually)
Everything in Free plus unlimited projects (tests, interviews, surveys, polls), unlimited panel and AI tester recruiting, 5 sharing channels, conditional logic, templates, 70+ language localization, SSO/SAML, custom API & webhooks, CRM inte
A research method where users complete real tasks while AI guides the experience. It captures screen activity, voice feedback, and behavior, then automatically analyzes the data to surface usability issues, patterns, and actionable insights — dramatically faster than traditional testing.
Per TheySaid's side-by-side studies, of all issues discovered 79% come from both human and synthetic audiences, 10% only from synthetic, and 11% only from humans — so synthetic-only tests uncover 89% of total issues vs 90% for human-only tests, and synthetic testers also catch issues humans miss.
Synthetic tests finish in minutes; tests with real people return in hours rather than weeks. Setup takes one person under 2 minutes, so you can run your first test right away.
Yes. The Free plan includes 5 AI user tests per month, 50 AI surveys per month, 1 workspace, 1 active test, and the AI project creator, moderator, and analytics — no credit card required.
AI design tools (Claude Design, v0, Bolt, Lovable), AI coding tools (Claude Code, Cursor, Replit), and ticketing tools (Jira, Linear, ClickUp, Monday.com). One person connects it once; nobody else needs a new login.
Yes — you can test with your own customers, testers recruited from TheySaid's panel, or third-party panels. TheySaid screens, schedules, and pays participants for you.
AI-generated summaries, task completion metrics and time-on-task stats, patterns of repeated issues, highlight video clips of real user behavior, and prioritized recommendations.
They start from a base model trained on 100,000+ hours of real usability sessions, then learn from your own customer data. TheySaid says it's the only technology where synthetic testers record their screen while speaking thoughts aloud, producing video deliverables rather than text walls.