Your application. A complete conversation layer.
Use REST, GraphQL, WebSocket streaming, SDKs and MCP to bring voice into your product. Keep your own agent brain, business logic and provider strategy while Voice handles the conversation experience.
Go from a test conversation to a versioned, observable integration.
A conversation, connected to your business.
Choose your interface
Start with the surface that matches your application and transport rather than rebuilding the same session plumbing.
- REST and GraphQL APIs
- Realtime WebSocket and WebRTC
- Python, TypeScript, Go, Ruby and client SDKs
Bring your own intelligence
Use your own prompts, tools, providers or agent brain while preserving coherent conversation and channel behavior.
- Custom tools and MCP
- BYOK and self-hosted inference
- External state and client-data contracts
Move safely into production
Make environments, versioning and observability part of the first integration.
- Sandbox and test conversations
- Short-lived embed tickets and plan-scoped keys
- Idempotency, pagination and session debugging
From a conversation
to a completed task.
Connect the steps that make the outcome possible.
- 01
Create your organization and key
Set up the environment and a scoped API key for the integration. Keep application secrets on the server.
- 02
Run a test conversation
Use an SDK, browser widget or streaming client and exercise interruption, handoff and recovery.
- 03
Connect a real business action
Define typed inputs, permissions and approval behavior, then inspect the external result.
- 04
Version, publish and observe
Release the agent configuration, connect the channel and follow events, traces, usage and outcomes.
Make it useful
in your business.
Start with a specific conversation, connect the right systems, and define what a successful outcome looks like.
Embed voice into a SaaS product
Build a branded client experience
Connect an existing agent brain
Use speech capabilities in a custom application
Can I use speech without building a receptionist?
Yes. The speech-to-text and text-to-speech capability paths serve builders who need individual speech functions. Voice product subscriptions and standalone Vagary capability contracts remain distinct even where underlying identity and infrastructure are shared.
Is there a way to test before live traffic?
Use the sandbox, test conversations, voice previews and session debugger to verify the integration. Keep Test and Live environments explicit, and review the test-conversation billing treatment for your plan.
Where are the SDK and API details?
The API reference and developer documentation cover interfaces, client setup, authentication and event contracts. Follow the reference for actual endpoint and SDK syntax rather than inferring requests from product illustrations.
Keep exploring.
Build voice into your product. Keep control of the experience.
Use APIs, realtime clients and embedded conversations to make voice part of your application. Bring your own agent brain or compose the speech, knowledge and action capabilities your product needs.
Explore productYour stack, connectedLet a conversation start the next piece of work.
Connect Voice to n8n, Make, Zapier, Slack, Microsoft Teams and your own HTTP endpoints. Carry intent into approved actions and events your operation can follow.
Explore integrationsGive every conversation
somewhere better to go.
Start with one agent. Build an operation around what works.