We are building a multi-LLM agent framework with dispatch via knowledge graph. Knowledge graphs involved are quite small, so we’d like to avoid the hassle of a separate process (currently using kuzu mainly because it’s embeddable).
Is it possible to run typedb inside a Python process, or do I always have to spin it up separately as per the README?
Sorry, we missed this. (I hope you came to discord and had better luck there)
TypeDB does indeed need to be run as a separate process. You’ll also need java installed since it’s written in java. There was talk about making it embeddable for robotics applications but it was abandoned as focus moved to writing TypeDB 3.0 in rust.
That should be easier to make embeddable, though I’m not aware of it being part of current plans.
Hey 
As per my knowledge, Typedb (formerly known as Grakn) is designed to run as a separate server process, so you’ll need to spin it up separately as per the README. While this might seem like extra overhead, it ensures that the database can manage its resources efficiently and maintain a consistent state.
I hope this will help you. I am happy to help you.
Respected community member 