AgentPhone gives AI agents their own phone number so they can handle voice and messages through one unified webhook. It is built for workflows where an autonomous agent needs a phone identity, including receptionists, AI workers, on-call engineering flows, and coding agents.
The service provisions US and Canadian numbers instantly through the API. Incoming voice calls are transcribed in real time and sent to a webhook as text; the agent replies with text, and AgentPhone handles the speech layer. It uses the same event shape for voice and messaging, and the page also mentions automatic conversation threading and native MCP support. The examples and FAQ reference use cases such as sales follow-up, delivery updates, 2FA and verification inboxes, and real-time conversational tasks.
Integration options include a Python SDK, a Node.js SDK, a REST API, and an MCP server. The page says it works with Claude Code, Cursor, OpenClaw, Windsurf, and any MCP client. The website also shows an NPM option and code examples for creating agents, buying numbers, and making calls. The page identifies Relay Innovations, Inc. and says the product is backed by Combinator.
In the VoIP, telephony & mobile networks space, langchain-agentphone takes a focused approach. It enables AI agents to interact with telephony systems via LangChain integration. langchain-agentphone is an open-source project aimed at AI developers building agentic telephony applications. The project is open source (MIT). langchain-agentphone is available on the web and the command line.
langchain-agentphone first shipped in 2026. Key capabilities include telephony integration, langChain support, and SMS tools. It exposes integrations via a public API and an MCP server.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do