Framer AI AgentsDesign and publish professional sites with AI
Promoted
Launching today
Write a post, a listing, a product or a headline, and 10,000 computed AI residents read it. The 600 it should matter to see it first, and it reaches the next 1,500 only if more readers were glad than annoyed, so a weak text dies for half a cent and a good one reaches everyone in 14 seconds. You get who stopped, liked, reposted or blocked, by interest, job, age, city and budget; the questions buyers would ask a listing first; and a demand curve over a product's price ladder. No sign-in.
Jevtown
Hi Product Hunt.
Most Jev demos ask the model for one decision: moderate this comment, route this email, score this ticket. I wanted to see what ten thousand decisions about the same text look like, so I built a town and put Jev in every house.
You write a post, a listing, a product or a headline. The 600 residents it should matter to read it first, and it travels further only while more of them are glad than annoyed. A weak text dies in the first wave for half a cent. A good one reaches all 10,000 in 14 seconds for about ten cents.
The test that convinced me: I wrote one iPhone listing two ways. The version that lets the buyer pay on inspection reached 2,100 residents and 142 wrote to the seller. The advance-payment-only rewrite reached 600 and stopped, with 204 of them suspecting a scam.
Three things I measured before building any of it, in case they help someone else building on Jev:
1. 200 personas in one request answer the same as one asked alone, so batching costs nothing in accuracy.
2. Reversing the order of the options shifts answers by 0.062, two and a half times the noise between two identical calls. So the order is fixed and never shuffled.
3. Asking "what is the highest price this buyer would pay" turns 90% of people into buyers. Writing the base rate into the question gives 48%, which matches what they actually do elsewhere in the same run. The calibration is real, but it calibrates the question you wrote.
No sign-in, no accounts. Ukrainian and English, and the language of your text picks the town.
Try it: https://jevtown.ivanhabor.com
The 30-second film of the two listings: https://youtu.be/Ktm2qwW7JAo
I would most like to hear about texts where the town got it wrong.
@ivangabor What kind of text has surprised you the most so far when the town gave an unexpected result?
Jevtown
@oliver_graf1 "Good morning everyone." I sent it through the first wave together with five other weak texts, spam and a scam listing among them, and expected all six to die there. Five did. "Good morning everyone" got through by a hair, which is about how such posts do in real feeds.
Do you also track how long they spent reading before they reacted.
Jevtown
@jenniferdavis No, Jevtown doesn't measure reading time. For each resident, Jev answers one question: what is the most this person does with the post? It returns probabilities for scrolling past, reading, liking, reposting, following and blocking. The closest thing to attention is the share who stopped instead of scrolling past.
Finally a tool that does not make me connect all my socials just to try it. How long does a typical test take to finish.
Jevtown
@jordantaylor58 Thanks! A text that doesn't land stops at the first wave of 600 residents in 2 to 4 seconds. One that reaches all 10,000 takes about 14 seconds.
Have you compared your AI reactions to real human reactions to see how close they are.
Jevtown
@harrywilson No, not against real audiences. I checked two things. When I made the same 40 residents gardeners, 93% of them stopped at a post about tomato seedlings. When I made them programmers, 15% did. And the rule that decides whether a text travels further separated all six weak texts I tried, including spam and a scam listing, from six normal ones.
You can compare Jev with one real person, yourself. At jevtown.ivanhabor.com/me you describe a resident, ideally you, and say what they'd do with 12 posts. Then you get 8 new posts. You answer first, then Jev answers for your resident without seeing those answers, and the page shows how many of its answers matched yours and how many would have matched from the description alone.
These are still simulated readers, so the numbers are best used to compare two versions of the same text.
Story Bowl
I like it. Beautiful UI and cool Jev usecase. I'm sure many people are using Jev for AI detection but seems like a good use case here too
Jevtown
@robbalian Thanks, glad you like the UI! An AI detector asks Jev about the text. Jevtown asks it what each of 10,000 residents would do with the post, and the answer depends on who they are. When the same 40 residents were gardeners, 93% stopped at a post about tomato seedlings. As programmers, 15% did.
The idea of 10000 readers reacting before publishing is such a good pitch. Is the audience configurable for a niche like tech founders only
Jevtown
@leomartin357 Thanks! Not on the site. Everyone posts to the same town of 10,000 residents, and about 800 of them are into startups. A post for founders reaches many of them first, because the first wave goes mostly to the people the text should matter to most. When the run finishes, you can filter the reactions by the startups interest and see what those residents did.
For a town of founders only, you'd need your own copy. The code is open under MIT, and the residents are generated in one file, public/shared/personas.js, where you can change the jobs and interests.
Jevtown
@leomartin357 Would you rather describe the founders yourself, or have the crowd built from your real followers?
Would love to see the questions buyers would ask part in action. Does it suggest rewrites too.
Jevtown
@shinyahayashi Here is one: https://jevtown.ivanhabor.com/p/171f5a1168
It's a Ukrainian iPhone listing written two ways. The first version gives the details and offers inspection on delivery. Of the 531 buyers who stopped at it, 24% would first ask "Is the price negotiable?" The second version says "prepayment only", and 43% of those who stopped would first ask for a safe deal or cash on delivery. It never got past the first wave, where 204 of 600 buyers smelled a scam.
It doesn't suggest rewrites, because Jev only answers questions and writes no text. You write the next version yourself, and the buyers' questions show what the listing is missing.