I am really surprised how useful the #Anagnorisis project has become for me over two years of development. Started as a local music player with an AI recommendation system that trains and adapts to my preferences, it has now expanded to visual and text modalities. While I haven't implemented everything I have in mind, I already use it daily for all my note-taking, listening to music, and managing my extensive personal photo archive.

I now have three instances running 24/7. One for me, for everything stated above. One for my kid, who can watch a well-maintained archive of cartoons instead of typical bs that YouTube’s algorithm constantly recommends kids to watch. And one for my mother for maintaining and labeling our family photos that go back almost a century. This is the kind of data I would be uneasy to trust to any cloud service out there.

I am still trying to figure out what to make of many different ideas that I have regarding the project, but my fuzzy goal is to make something like a personal, local Google for all your household data. Where a single tech-savvy person in the family can set up an Anagnorisis server, and everyone else can enjoy their shared data together. Keeping your important data close to the ground and not sharing it with any third party, while having all the main benefits of the search and recommendation services we're so used to.

I think one of the most important observations of mine is that recommendations are just another filter that we apply to our data along with many others that are possible. So it all comes down to data preservation and search. But now with complete control over the data and no need for any additional trust.

You can check out the open-source project here: https://github.com/volotat/Anagnorisis

#opensource #selfhosted #privacy #PersonalAI #DataOwnership #FamilyTech #photomanagement #decentralized