OpenHome - AI Smart Speaker Platform

The Problem

OpenHome is an open-source AI smart speaker platform that lets developers build custom voice agents. They needed two new abilities added: a WhatsApp Voice Sender module and an Alarm/Timer system. Both had to integrate cleanly with their existing voice pipeline architecture and ship within the community's active release cycle — meaning the code had to meet open-source contribution standards, not just work in isolation.

Our Approach

Built each ability as a standalone module that hooks into OpenHome's voice pipeline without modifying the core SDK. The WhatsApp Voice Sender captures voice input through Deepgram, processes intent through the LLM layer, and delivers voice note responses via the WhatsApp API — all within the existing ability interface. The Alarm/Timer ability parses natural language time expressions and queues execution events. Both were developed in collaboration with the CTO directly through the project's Discord community and shipped via pull request into the open-source repo.

Pipeline Breakdown