Welcome to VIP Discussions - an open standard for voice interaction

[raaj-love-to-code]

Visual interfaces have the DOM. Accessibility has ARIA. Voice has nothing.

Every voice integration today is built from scratch - tightly coupled to a
specific AI provider, a specific SDK, a specific platform. There is no shared
layer. No standard vocabulary between an application and a voice agent. No

VIP is an attempt to fix that.

The Voice Interaction Protocol defines a standard interface for voice-driven
applications: how a session is established, how application state is exposed
to an AI agent, how actions are declared and safely invoked, and how the
interaction state machine is synchronized between client and runtime.

Model-agnostic. Platform-agnostic. Device-agnostic.

---

The v0.1 draft is live and covers the core of the protocol - the state
machine, Voice Interaction Tree, Action Registry, interaction primitives,
and flow management. Message format schemas and transport layer spec are next.

Read it → https://voiceinteractionprotocol.io/docs/spec/introduction

---

This is an open effort. A standard only works if it is built with the
people who will implement it.

• Feedback on the spec → Ideas
• Questions about the protocol → Q&A
• Building something with voice → Show and Tell
• Everything else → General

If you're here, say hi.

-- Raaj