Re|InventAI Labs
Project · Voice agents

ReInvent Voice

Open-source voice-agent infrastructure for restaurant ordering workflows.

Restaurants and small service businesses need voice automation that fits into existing workflows without forcing them into a closed platform or generic chatbot interface.

GitHubDocsDemo
Voice processing
State: idle

Faint standby glow — awaiting input

Problem

Restaurants and small service businesses need voice automation that fits into existing workflows without forcing them into a closed platform or generic chatbot interface.

Solution

ReInvent Voice provides an open-source voice-agent engine with API-first design, workflow state management, menu/order understanding, optional UI components, and integration-ready architecture.

System overview
Voice ingest

Real-time audio capture via WebRTC and streaming ASR.

Workflow engine

Stateful conversation graph with menu/order intents.

Integration layer

Webhooks, REST APIs, and SDK-friendly outputs.

API concepts
POST/v1/conversationsCreate a new voice conversation
POST/v1/conversations/{id}/messageAppend a user turn
POST/v1/orders/parseParse spoken order into structured items
GET/v1/orders/{id}Retrieve an order state
POST/v1/webhooks/order.createdWebhook fired on order creation
Example restaurant workflow
STEP 01
Caller dials in
STEP 02
Voice agent parses menu intent
STEP 03
Order is structured & confirmed
STEP 04
Webhook fires to POS / kitchen
Developer integration
Drop-in API

Plug ReInvent Voice into your existing backend through REST and webhook surfaces — no UI lock-in.

Optional UI kit

Geometric voice-state components you can render in your own admin dashboard or customer interface.

Roadmap
v0Prototype voice loop + workflow states
v0.1Order parsing + structured menu schema
v0.2Webhook integrations + reference UI
v1Evaluation harness + deployment kit

Explore the other ReInvent systems

All projectsReInvent OpsReInvent Docs