Paste the name of a public Telegram channel and get a card for its last year. Jev, TypeSafe's decision model, reads up to 1,500 posts one by one and answers four questions about each: which of ten kinds of post it is, whether it is a paid ad, whether it uses clickbait, whether it pushes on fear or anger. The card shows the mix month by month and links the posts that scored highest, so you can check the model yourself. No sign-in, no Telegram account. Open source.
Hi Product Hunt! I found one more use for Jev, the decision model from TypeSafe, and built a small project with it: an X-ray for Telegram channels.
Jev writes no text. It takes a post and four typed questions and returns a probability for each answer, in about 300 ms. At that speed and price you can read a whole year of a channel post by post instead of asking a chat model for a summary.
What you get: the share of news, opinion, ads, jokes, the author's own promotion and five other kinds, three percentages for ads, clickbait and emotional pressure, and the mix for each of the last twelve months. Durov's channel took 6 seconds and $0.0043: 53% of it is about his own projects.
Jev makes mistakes. On a channel with no paid advertising it still marked two partner promotions as ads. So below every card the page lists the posts it scored highest, with links, and you can check them yourself.
It is open source (MIT), one Cloudflare Worker on the free plan, and runs on your own TypeSafe or OpenRouter key.
Try it: https://wrapped.ivanhabor.com
Code: https://github.com/gaborishka/je...
If a card looks wrong for a channel you know, tell me here. I want to see where the model misses.
Jevtown