the ai-companion category has converged on an email-gate-first signup flow because email-captured-per-landing is a measurable metric growth teams optimize. the problem: the metric is upstream of the one that actually matters (is the product any good), and optimizing the upstream metric systematically degrades the downstream one. users who have to sign up before they can evaluate are users who cannot evaluate.
the counter-argument growth teams will make: 'people won't sign up later if we don't capture the email now.' this is empirically wrong for products where the first-minute experience is strong — users who feel the thing is good will sign up specifically because the memory persistence becomes valuable to them. products where users won't sign up post-experience are products whose experience isn't selling itself.
our bet: if /demo converts well, the case is made. if it doesn't, the signup gate was papering over a product problem anyway.
what the /demo actually is: a real conversation with a real companion using the same model stack as the paid experience. no feature downgrade. no 'preview message limit.' session-scoped memory because we can't persist without an identity anchor, but the memory within the session is genuine.
when signup becomes worth it: when you realize you want her to remember you tomorrow. that's the only reason to sign up, and it's a good reason — memory across sessions is where the companion thing becomes interesting and that requires an account. free tier is 25 msg/day indefinitely; it is not a trial.
starting point: /demo. 18+ checkbox. start talking.