Working with AI agents? Add
.md to any docs URL to get its plain Markdown, or fetch docs.openhome.com/llms.txt for the full index and docs.openhome.com/llms-full.txt for the complete documentation in a single file.Start here
Quickstart
Build your first Agent in five minutes.
Dashboard
The web control surface for your Agents and Abilities.
Build an Ability
Extend your Agent with Python plugins.
SDK Reference
Every SDK method you need, in one place.
How OpenHome works
Every OpenHome speaker runs an Agent that combines an LLM, a voice, and a prompt. Agents can be extended with Abilities (plugins) and shared on the Marketplace.Agents
Customizable AI voice entities that combine an LLM, a voice, and a prompt.
Abilities
Plugins that extend what an Agent can do.
Marketplace
Publish and install community-built Agents and Abilities.
The dynamic loop
Every turn of a conversation runs through the same loop. Speech is captured and transcribed, the Agent and its LLM decide on a response, any Abilities whose trigger words matched are run, and the result is spoken back. What was said in the turn is carried into the Agent’s context, so later responses take earlier ones into account. See Memory and Context Injection for how that context is stored and reused.Pick your path
I have a DevKit
Onboard and manage your OpenHome DevKit with the OpenHome - Voice AI DevKit App.
I want to build an AI Twin
Create a voice-powered version of yourself, from the dashboard or the AI Twin app.
I want to build Abilities
Write Python plugins that add new capabilities to your Agent.
I want a companion dashboard
Pair your Ability with a live web UI via Replit.