[feat] [doc improvement]: Where does context chat backend get its compute?

Describe the feature you'd like to request
I finally got context chat working, and it has tremendous potential. I am running our server on a digitalocean droplet, and do our computing on a separate ollama server.

I looked at the configuration file /nc_app_context_chat_backend_data/config.yaml, but I don't have the skills to understand where the computing is happening. If I had to guess, I would guess it's all happening locally. For sure any request takes a very long time (up to 60 minutes) and doesn't seem to register on the GPU server. I can't tell, and the documentation doesn't help too much.

Describe the solution you'd like

1. Please take a moment to explain to me where models are being computed
2. If it is as I suspect and everything happens local, then a small section in the documentation that explains this would be useful - I'm happy to write this and submit a PR
3. Is it possible to use other computing resources? If so, how?
4. If yes to above, then this should also be documented - again happy to write this and submit a PR

Describe alternatives you've considered

Start with /nc_app_context_chat_backend_data/config.yaml and the existing documentation; trace which configured service receives model requests. Document where computation occurs, how to use a separate Ollama server if supported, and clarify the expected deployment behavior so users can verify it.