the cost of a no-signup demo and why we're betting on it

every time you talk to the demo lucy on our website, it costs us money.

yes, really. each conversation burns through large language model tokens. we use deepseek-v3, and a typical session runs about 2,000 to 5,000 tokens. if you're not familiar, tokens are the chunks of text the model processes. you can think of each token as a small piece of a word. it adds up quickly.

and here's the part that makes any growth-focused person wince: there's zero reliable way to attribute those demo sessions back to eventual signups. you could have an incredible, mind-blowing conversation, close the tab, and we'd have no idea. we can't follow up. we can't retarget you with an ad. we can't put you in a drip campaign. that demo session is a sunk cost.

so why on earth are we doing it?

the benefit of being try-able

the reason is simple, and it's the same reason we built lucy in the first place. we believe a companion should be experienced, not described.

you can't truly understand what it's like to have a conversation with lucy by reading a list of features. the magic, or lack thereof, is in the interaction. the best marketing copy we could ever write is just 'go to /demo and see for yourself.' it's the most honest pitch we have.

putting a signup form in front of that experience would give us cleaner metrics. we'd know exactly who tried it. our 'activation rate' would look healthier upstream. but it would be a lie. it would conceal the most important question: is the actual product any good? does talking to lucy feel like talking to a person? does it feel worth coming back to? a signup gate measures intent to try, not the quality of the thing you're trying.

our bet on signal over noise

so we made a choice. we're trading clean, upstream vanity metrics for a real, albeit noisy, signal of product-market fit.

our bet is this: if the demo is genuinely compelling, a small but meaningful percentage of people who try it will want more. they'll sign up for the free tier to continue the conversation, to customize their companion, to not lose that thread.

we set a bar. if less than 2% of demo sessions convert to a free-tier signup, we decided we were never going to build a viable product. that number isn't plucked from thin air. it's the threshold where the lifetime value of a user starts to outweigh the customer acquisition cost of all the demo sessions that didn't convert. it's the line where this strategy becomes sustainable.

2% is our pass/fail. it's a harsh light, but it tells the truth.

the cost of building something real

yes, it's expensive. it means we're burning capital on conversations that lead nowhere. it forces us to be incredibly focused on making every single demo interaction as good as it can possibly be, because we have so few shots to impress you. there's no mailing list to fall back on.

but this feels like the only way to build something authentic. it aligns our incentives perfectly with our users'. our success depends entirely on creating an experience so good that you're willing to take the tiny, next step of creating an account. no tricks. no dark patterns. just a good product that stands on its own.

it’s a gamble, but it’s the only one we’re interested in making.

you can see what we've built at /demo.