Skip to content

COOORN/AssistGPT

Repository files navigation

A GPT client that has long term memory.

How it works

Every time a message is sent, embeddings are created and are relevent documents are retrieved(shown in the context card). There is also a persistent memory that the AI always has access to

Full privacy: all completely local using localForage(an indexedDB wrapper).

To - do not in any order:

  • Web browsing
  • Local AI - Once they make LangChain JS/TS support better, this should be possible