In bluesky you can create news feeds, that will curate a list of posts based on whatever search criteria you give. You basically host an API and when somebody wants to view a feed, the api returns a list of posts that match the search criteria.
The current state is pretty bad. there are about the 3 hosted feed creators. I'm using one of the early ones where you create a JSON file and have primitives like:
- Include posts from this list of users
- Include posts that have this regex
- include posts that have these hash tags
- sort by date/Hacker rank
There is a newer feature that allows the algorithm to see what has been liked in the feed to give a particular post more reach or higher placement.
What Kagi could do would be to make this simple and would be a combination of kagi search and maybe AI. It would allow a user to create a feed based off of search terms and what they have liked. Esentially Kagi could be (one of) the algorithms that chooses what people see in bluesky.
This is a screenshot of one of the current feed builders, where it is a visual tool to create a JSON doc that powers a feed api:
A user would create a Kagi powered Feed using normal english search terms:
I want to see picture of sheep, not cartoons, but sheep on a farm. Also include black and white landscape photos. Do not include any political posts, or anime, or adult content.
The user would get an initial feed of pictures, and by liking posts in the feed would build up a profile of what they like.
