Stunt Double vs TheySaid
TheySaid is a user testing platform with surveys, AI-moderated interviews and usability tests. Since its 4.0 release, a test can use synthetic testers (TheySaid calls them synthetic users), trained on your customer data from a base model built on more than 100,000 hours of real usability sessions, or real people: your own customers, recruited testers or third-party panels, recorded with screen and voice as they think aloud. It runs from Claude Design, v0, Replit, Bolt and Lovable, from Claude Code and Cursor over MCP, and when a Jira ticket finishes design.
Side by side
Every fact about TheySaid links to the public page it came from, checked on November 17, 2026. Pricing and features change, so check their site before you decide.
What it tests
- TheySaid
- Usability tests with synthetic testers or real people, surveys, AI-moderated interviews, polls and forms, and a spec check of the working build before the pull request merges.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Checklist runs (pass or fail per requirement, with a screenshot and the reasoning), interviews with actors, design reviews against your guidelines, and triaged feedback.
Real browser on live sites and preview deployments
- TheySaid
- Yes, from any URL you provide: staging environments, beta versions and live websites. Some sites need a browser extension. Preview deployments are not described.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Yes. An actor drives a real browser on any URL, including a preview deployment. Public URLs work on every plan; Startup and up add private URLs, signed-in credentials and an email inbox.
Figma prototypes
- TheySaid
- Yes. Tests run on Figma prototypes from their URL.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Yes. Actors go through Figma prototypes, and design reviews run on Figma frames, from the app or the Figma plugin.
Who does the testing
- TheySaid
- Both: synthetic testers trained on your customer data, and real people (your own customers, recruited testers or third-party panels). Sessions record screen and voice as the tester thinks aloud.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Actors: AI users you cast with a role and a goal. They use the product and report what they did, step by step. Behavior, not opinions.
MCP server
- TheySaid
- Yes. Connect once over MCP and run synthetic user tests from Claude Desktop, Claude Code, the API or any MCP client.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Yes, a hosted MCP server on every plan, Free included. It creates and runs checklists and interviews, and reads the results.
Pull request comments
- TheySaid
- Not confirmed from their public pages
- Stunt Double
- Yes. With GitHub connected, a checklist runs on a pull request’s preview deployment and posts its results on the pull request, and workflows can start on GitHub events.
Pricing model and entry price
- TheySaid
- Free plan at $0/mo, no card: 5 AI user tests and 50 AI surveys a month, one workspace and one active test. Enterprise priced on request, billed annually.Source, November 17, 2026 (opens in new tab)
- Stunt Double
- Free plan for one person: two actors and 60 minutes of runs a month, no card. Solo Trial $9/mo for one person, up to three months; Hobby $79/mo for a team of three.
Choose TheySaid when
- You want tests with synthetic testers and with real people, recruited or your own customers, on one platform.
- You want tests to start from your design tool, your coding agent or a Jira ticket, and to test iOS and Android apps too.
- You run surveys, AI-moderated interviews and polls alongside usability tests.
Choose Stunt Double when
- You want requirements written in plain language and checked pass or fail, with a screenshot and the actor’s reasoning for each.
- You want a checklist to run on each pull request’s preview deployment and post its results on the pull request.
- Your flows need a signed-in account, an email inbox or an SMS code to get through (Startup plan and up).
Sources
Public pages from TheySaid, read on November 17, 2026. If something here is out of date, tell us and we will correct it.
- TheySaid homepage (opens in new tab)https://www.theysaid.io/
- TheySaid pricing (opens in new tab)https://www.theysaid.io/pricing
- TheySaid: AI user testing (opens in new tab)https://www.theysaid.io/use-cases/ai-user-testing
- TheySaid 4.0 (opens in new tab)https://www.theysaid.io/v4
Keep reading
- Test apps and prototypes built with AIActors go through a prototype or an AI-built app before anyone else does.
- Synthetic researchWhat a study with actors answers, and the questions that still need people.
Other comparisons
- Stunt Double vs VersiveAI research platform: persona tests on prototypes and sites, plus AI-moderated studies with people
- Stunt Double vs MazeUser research platform: prototype and website tests, surveys and interviews with recruited people
- Stunt Double vs TestSpriteAI testing agent: generates and runs end-to-end and API tests from your IDE or CI
- Stunt Double vs MomenticAI end-to-end testing: plain-English tests in your repo, run in CI on web and mobile
- Stunt Double vs Synthetic UsersAI participants for interviews, concept tests and website tests, grounded in your audience data
Get started
See it on your own product
The Free plan is one person, two actors and 60 minutes of runs a month, with no card. Point an actor at a URL and read what it did.