Talqing is a platform for building AI agents that talk to people on the phone, in a browser and in chat, then publishing them and reading back every conversation. An agent is a system prompt, a model stack and the tools it may call; anything you could write by hand as a LiveKit agent you can build here.

​
The three channels

An agent runs on exactly one channel, which decides its model slots. An agent task is an agent nobody talks to: named inputs in, typed output out. Use one for research, classification and drafting.

​
Bring your own keys

Talqing holds no provider credentials. Every model runs on your OpenAI, Deepgram, ElevenLabs, Gemini, xAI or other account, billed to you by that provider. Nothing runs until you add a key for every provider your agent uses: publishing is refused, naming the missing provider. See provider keys. Talqing charges one platform fee from prepaid credit: $0.0035 per voice minute, $0.01 per video minute, $0.0001 per answered text message. See pricing and credits.

​
Four ways to build

All four reach the same operations, one per API endpoint, acting as you with your role.
  • The dashboard at https://app.talqing.com: editors for agents, tools and tasks, plus telephony, conversations and observability.
  • The API and SDKs: https://api.talqing.com/v1, with TypeScript and Python clients and a React layer for browser calls.
  • The CoPilots: a build assistant in the agent, tool and task editors.
  • Your coding agent: point Claude Code, Codex or any MCP client at our MCP server.

​
What you can build

​
Start here

Read core concepts for the vocabulary, then pick a quickstart.

Build a voice agent

Dashboard only: add keys, build an agent, talk to it in the browser.

Take a real phone call

Connect your carrier, import a number, point an agent at it.

Talk to a text agent in code

One script in TypeScript or Python.

Build from your terminal

Point Claude Code or Codex at the MCP server.