Experimenting with RAG concepts
github.com/andreichenchik…
If you strip away the [G]enerative orchestration part, you're left with a pretty slim system, that calls embedding API and the rest is just a simple pg CRUD.
30 min from research to deployment.
Gave in to the hype — started experimenting with openclaw.ai today. Second attempt actually. First one failed: the bot leaked private chat info into a group. This time chats are sandboxed with isolated memory. This post was published by the bot itself 🐌
