Skip to content
  1. Home
  2. AI tools
  3. CAAL

CAAL

Local voice assistant whose tools are n8n workflows exposed over MCP

What it is

A self-hosted voice assistant built on LiveKit Agents. Its abilities are n8n workflows discovered through MCP, so API credentials stay in n8n and the model only sends parameters to webhooks. It ships with a fine-tuned 8B model for tool calling and a registry of reviewed tools. For home-lab users who want a private assistant they can extend.

Good for

  • Control smart-home devices by voice
  • Add a new voice command by building an n8n workflow
  • Keep API keys away from the language model

What you need

Docker, an n8n instance and a local model runtime such as Ollama

Worth knowingRuns speech and LLM models on your own hardware, so it suits tinkerers with a capable machine; last pushed May 2026.

Sources

The summary is ours, written from the README. Spotted an error? See how we work.

More for productivity & knowledge

See the topic →
  • Ollama

    ollama/ollama

    Download and run open LLMs locally behind a simple CLI and API

    ToolGitHub stars: 187k
  • Open WebUI

    open-webui/open-webui

    Self-hosted chat interface for Ollama and OpenAI-compatible APIs

    ToolGitHub stars: 159k
  • llama.cpp

    ggml-org/llama.cpp

    C/C++ LLM inference engine with CLI and OpenAI-compatible server

    ToolGitHub stars: 135k
  • Claude-Mem

    thedotmack/claude-mem

    Records what your agent did and loads a summary into later sessions

    ToolGitHub stars: 117k