That sounds like a pretty cool use of FUSE. I keep trying to use it for remote systems, but also keep getting into the how to set it up pain.
I'm wondering if there's a potential for integration into a codebase for keeping a knowledge base updated, where there's a mix of developers, support staff, product, and of course, AI Agents?
It’s super easy for Linux but Apple really hate developers. FSKit is super half baked and there are very few apps I’d suffer a reboot for. So. If a read-only knowledge base appeals, try it out. Otherwise if actually keeping files on disc is no obstacle, the web sockets api works just fine
I have another project where I was doing semantic search via standard RAG and realized an agent plus grep was actually way better than matching user text to obscure legalese vectors. Claude already knows CFR from its training data so targeted grep recalls data with less ranking gymnastics.
That sounds like a pretty cool use of FUSE. I keep trying to use it for remote systems, but also keep getting into the how to set it up pain.
I'm wondering if there's a potential for integration into a codebase for keeping a knowledge base updated, where there's a mix of developers, support staff, product, and of course, AI Agents?
It’s super easy for Linux but Apple really hate developers. FSKit is super half baked and there are very few apps I’d suffer a reboot for. So. If a read-only knowledge base appeals, try it out. Otherwise if actually keeping files on disc is no obstacle, the web sockets api works just fine
I've been wondering how knowledge bases and documents can evolve in agentic era. This seems promising.
I have another project where I was doing semantic search via standard RAG and realized an agent plus grep was actually way better than matching user text to obscure legalese vectors. Claude already knows CFR from its training data so targeted grep recalls data with less ranking gymnastics.
[dead]
[flagged]
[dead]