AnythingLLM connects to Ollama as a local LLM provider for chat and embeddings, letting users run private document chat and agents without sending prompts to a hosted model API.