Skip to main content
Kataven is the platform behind production AI voice agents. Sign up, point a phone number at an agent, and you have a 24/7 receptionist that books appointments, answers FAQs, and runs outbound campaigns — speaking real-time over PSTN, WebRTC, and embeddable web widgets. The Hub API is how you drive it programmatically: every screen in the dashboard at hub.kataven.ai is a thin client over the same REST endpoints your scripts, CI pipelines, and AI tools (Claude, ChatGPT, Codex) call.

​
What you can build

AI phone agents

Inbound + outbound voice over Twilio, Plivo, or Vobiz. Sub-800ms latency end-to-end.

Bulk outbound campaigns

Upload a CSV, dial thousands of contacts with cap-aware concurrency, watch live metrics over SSE.

Embeddable widget

Drop a launcher into any website. Anonymous or authenticated visitors. Per-domain allowlists.

Tools, knowledge, integrations

Built-in + custom tools. Stripe, Shopify, Slack integrations. Encrypted credentials.

Drive everything from AI

Connect Claude Code, Claude Desktop, ChatGPT, Codex, or any MCP-aware client. They call the API for you.

Multi-tenant by default

Per-account Postgres isolation, per-tenant rate caps, no shared state.

​
Pick your interface

​
Start in 60 seconds

1

Get an API key

Open hub.kataven.ai/settings, click the API Keys tab, then Create API key. Pick an expiry (Never / 1y / 90d / 30d), click Create, and copy the sk_live_… value — it’s shown once.
The plaintext key is only displayed at creation time. We store an HMAC, not the token. Lose it and you mint a new one.
2

Make your first API call

Set the env var, then list your agents in five lines of code. Full walkthrough at the Quickstart.
3

Build your first voice agent

Walk through Your first agent — create the agent, attach a phone number, and place a test call. ~10 minutes from blank account to ringing your cell.
4

Hand it off to AI

Once your agent works, let an AI tool drive it for you: Claude Code (or Claude Desktop, ChatGPT, Codex) can manage agents, run campaigns, and place test calls without you writing code.

​
What’s in the docs

  • Concepts — what every primitive (agents, tools, playbooks, FAQs, campaigns, widget, telephony, multi-tenancy, rate limits) actually does.
  • Guides — opinionated walkthroughs for common end-to-end flows.
  • API Reference — every endpoint, every field, with try-it.
  • Python SDK + Node SDK + MCP Server — language- and tool-specific references.

​
Help

  • Found a docs bug, hit a confusing error, or want an API change? Reach out at support@kataven.ai. Include the doc URL or the API endpoint you were on — we’ll route it.