I'm a fan of Techmeme, which clusters tech news by topic, so I decided to do the same with edtech and help me keep up with edtech. But it helped only a little. Unlike in tech, where daily news is usually heavily clusterable because people comment on daily trending topics like apple's latest launch etc., in edtech, we don't really have these trending topics often, but we do have these persistent ongoing beats like AI in education or Edtech policy etc. So I decided to leverage the help of the newly released Jev model (a model that's super fast and cheap and works great with categorization tasks) to help categorize these items into different topics and tags, which seems to work really great so far (and has kept the cost low).

In terms of implementation, it's basically a pipeline that processes a list of sources consisting of trade media and indie newsletters and blogs with RSS feeds, and tags them with Jev using a hierarchy of topics related to edtech. Users can follow and pin individual topics, like "higher ed", "Europe" etc., so we don't need any persistent profiles for personalization; preferences live in the browser. Articles are clustered using cloudflare's vector store, and related articles are also recommended with that.

I have no doubt this will work even better with tech news, but there are plenty of alternatives out there, so I'm not going to indulge myself further. My takeaway is that Jev for content tagging is great so far. And edtech is kinda fun to keep up with.

The project is called EdFountain at https://edfountain.com

The goal is to create the water fountain for edtech.

For those who like to follow edtech, maybe it's useful to you as well. Feel free to drop a coin and make a wish on the /about page if you have other source recommendations. :D