Compare

nth Labs and Synthetic Users

Both put model-made respondents in front of your idea. They differ in what stands behind a respondent, what the respondent reads, and whether the run grades itself.

What Synthetic Users does

Synthetic Users describes itself as user research at the speed of AI: you specify the participant you want in as much detail as you need, and its agents, given personality profiles, sit for interviews and concept tests. Its pricing page lists plans from $12,500 a year, drawn down as a token pool per interview.

What nth Labs does

Every seat on the panel is a census record drawn from the market you declared, with a real survey respondent matched to it from Columbia Business School’s Twin-2K-500 study and up to a hundred of that person’s own answers shown to the model. The panel does not read a description of your product; it reads the map a real browser drew of it an hour earlier, screen by screen, and the pages as the browser digested them. Every run asks a sample of seats questions their real people already answered and prints the match as a share of the human ceiling; under a floor the money model refuses the panel’s rates. The whole panel is priced before it runs, and the price is a run, not a year.

Which to pick

  • You want to interview a described persona about a concept, at length, before anything is built: Synthetic Users is built for that.
  • You have a product a browser can open and you want to know what breaks, who quits and where, with the panel’s own accuracy printed beside the answer: nth Labs is built for that.
  • Neither is a replacement for real people. Both are a first pass before them.