Tired of re-explaining yourself to AI every single session?
Your LLM forgets decisions, goals, and constraints the moment the chat ends.
I built knowledge-worker โ a local-first personal knowledge graph that gives your AI durable, source-backed memory.
Thread๐
Built on these principles:
โข Provenance first every claim traces back to the exact source excerpt
โข Local first your graph stays on your machine (no cloud)
โข Review before merge human oversight + deterministic validation
Works great with Ollama for local workflows.
How it works:
1. Ingest your Markdown notes
2. LLM proposes concepts, decisions, goals & relationships
3. You review/approve every claim (with source excerpts)
4. It merges into a local JSON graph with full provenance
No more โas I mentioned earlierโฆโ hallucinations.
@merlindotcom_@cursor_ai@Paperclip_AI Everytime I look at your posts Iโm like how much better can this get and you get me again in the morning ๐๐๐ญ๐ญ
Seton hall student section really blows my mind, like do they not understand the players on the court feed off the energy. They need to do better. Especially if we are going to win our remaining games .
Iโve never seen such outrageous calls at this seton hall game. Iโve said the players were investigated now itโs time for X to do there thing with these refs .