Alexander Priymak · Budapest

← priymak.com

Another idea · SameVibe · intent-based social network

Find people who think, feel and dream like you.

Social networks today connect you to people you already know or to content. HR sites and dating apps make you fit a form. SameVibe is a new take on the social network, built around intent: targeting, the way ads find their audience, but pointed at people. Write one honest request, as unusual as you are, and it finds the people whose requests are closest to yours.

Request

Free text about yourself and who you are looking for: interests, values, odd habits.

Match

AI turns every request into a vector and shows the closest people live, with an insight on what you share.

Privacy

Public requests appear in Explore. Private ones are read only by the matcher.

Prototype of the MVP

Sample people. Here matching runs in the browser on topics; the product uses vector embeddings. The logo resets the prototype.

samevibe.app

How it works

  • 1. Write a requestInterests, values, odd habits and who you want to meet, in your own words. "Optimize text with AI" helps if you get stuck.
  • 2. Choose public or privateBoth go into the shared base. Public requests can be read in Explore; private ones are read only by the matcher.
  • 3. See matches liveEvery request becomes a vector; the closest people appear on the right, best on top, with an AI insight on what you share.
  • 4. Save and connectSaved people open as profiles with verified messengers, a website and an About Me written apart from any request.

Kept simple for the MVP

  • Free text, not forms"I walk in the rain, collect strange stones and dream of a book cafe in the mountains" is a valid search.
  • Privacy where it mattersPrivate requests are never shown; matches see only what you have in common.
  • AI explains each matchSo people know what to start the conversation with.
  • Messengers, not feedsContacts are messengers connected by signing in: verified, private, made for talking. No public feed to scroll.
  • Planned stackOpenAI embeddings for requests and profiles, a Pinecone vector index for matching.

Design explorations

Interface designs for the MVP. Earlier working names: Reunion, Soulmesh.