Use Local AI in BCCLI

Once a local runtime is running and its provider is registered, its models sit in the same list as the cloud ones. BCCLI can use Ollama — local or cloud — LM Studio, and any other OpenAI-compatible provider without a separate mode.

Check the list

bccli models

Each entry is provider/model. A local model appears when its runtime is running and has the model available; the same list also shows BotConnector Cloud models if you are signed in to it.

Pin a local model

bccli -m provider/model

Inside a session, /provider and /model switch without restarting, which is the fast way to compare a local model against a cloud one on the same task.

What changes when the model is local

The request never leaves the machine, so the answer is available offline — as long as the runtime and the model file are already in place. What you give up is the catalogue: a local model is whatever you downloaded, at the speed this machine can run it.

Conversely, if a local call fails while the cloud works, the runtime is the place to look, not BCCLI. Check it is running on the loopback address the CLI expects.

Next

• Local AI provider setup — if the provider is not in the list.
• Offline use — what still works without a network.