On the whole, it orchestrates a variety of models (TTS, OCR, LLM, etc.) and other tools (search engines, file storage) to make a configurable, private AI gateway. You can plug in your own third party API keys or self-host the models as well using Ollama.