Last updated: 2026-05-28
AgentPhone gives AI agents US or Canadian phone numbers via 1 API call, with unified voice and SMS webhook, real-time transcription, and MCP support. YC S26.
AgentPhone is a developer API launched in 2026 by Y Combinator (Spring batch) that gives AI agents their own US or Canadian phone numbers via a single API call. It handles voice and SMS through one unified webhook, with real-time transcription and MCP support for Claude Code, Cursor, and Windsurf. Pricing is not publicly listed; contact sales. Best for backend engineers building autonomous agent telephony workflows.
AgentPhone is a developer API launched in Spring 2026 by Meet Modi and Manav Modi as part of Y Combinator's Spring 2026 cohort. The platform solves a specific gap in the AI agent stack: AI agents that can handle web tasks lack the ability to make and receive phone calls or text messages, which are how most businesses still communicate. AgentPhone gives any AI agent its own US or Canadian phone number provisioned via a single API call, with voice and SMS covered by one unified webhook. The core architecture handles both voice and SMS with an identical event shape, so developers write one webhook handler for all communication types. Incoming calls are transcribed in real time and delivered as text to the webhook; the agent replies with text, and AgentPhone converts the response to speech. Agent code never needs to handle audio processing, PSTN connectivity, or carrier compliance directly. The platform ships with a native MCP server, letting developers integrate directly from Claude Code, Cursor, Windsurf, Zed, and any generic MCP client without additional integration work. AgentPhone targets backend engineers and AI agent developers building applications where agents need to contact people by phone. Specific use cases include AI receptionists that handle inbound business calls, sales agents that follow up with leads by text, on-call engineering workflows where agents page engineers via SMS, coding agents that receive verification codes, and 2FA distribution systems. Early adopters include teams at Google's Agent Development Kit, Replit, LangChain, and Alchemy. Pricing is usage-based and not publicly listed; the team handles rates on request. Phone numbers are currently limited to US and Canadian numbers, with international coverage in progress. The platform supports Python and Node.js SDKs and a REST API. Area code selection is available during provisioning.
Usage-based pricing, not publicly listed as of May 2026. Contact sales for rates. Phone number provisioning, voice minutes, and SMS usage billed separately. No confirmed free tier.
AgentPhone is a developer API that gives AI agents their own US or Canadian phone numbers, launched in 2026 by Meet Modi and Manav Modi as part of Y Combinator's Spring 2026 cohort. The platform solves a core gap in the AI agent stack: agents that operate online have no way to call or text real phone numbers without a telephony provider. AgentPhone provisions a phone number via a single API call and routes all incoming voice calls and SMS messages to a unified webhook. Voice calls are transcribed to text in real time so agent code only processes plain text, never raw audio. Early adopters include teams at Google's Agent Development Kit, Replit, LangChain, and Alchemy.
AgentPhone does not publish its pricing publicly as of May 2026. The platform uses usage-based billing and requires direct contact with the team to get rates for phone number provisioning, voice minutes, and SMS usage. There is no confirmed free tier. Developers evaluating AgentPhone should factor in US A2P 10DLC registration fees if they plan to send business SMS at scale, which adds $4-25 per brand and $10 per campaign plus carrier surcharges of roughly $0.003 per message on top of any platform rate. Contact the team at agentphone.ai for current pricing.
AgentPhone's core feature is one-call phone number provisioning: a single API call returns a US or Canadian phone number with optional area code selection. All incoming calls are transcribed to text in real time and delivered via webhook alongside SMS messages, using an identical event shape for both. The platform includes a native MCP server compatible with Claude Code, Cursor, Windsurf, Zed, and generic MCP clients, letting developers integrate without writing custom API wrappers. Official SDKs are available for Python and Node.js, plus a REST API for any other runtime.
AgentPhone does not have a confirmed free tier as of May 2026. The company is a 2-person YC-backed startup in early commercial deployment and handles pricing directly. There is no indication of a free plan or free trial on the website. Developers looking for a free alternative to evaluate phone number APIs should consider Twilio, which offers a free trial number with limited credits, or Vapi, which has a documented free tier for development use.
The closest alternatives to AgentPhone in 2026 are Retell AI, Vapi, and Twilio, each suited to different buyer profiles. Retell AI is a full no-code voice agent builder with call analytics and CRM integrations, better for non-technical teams needing a managed solution. Vapi is an API-first voice platform with public per-minute pricing and a free developer tier, making it easier to evaluate before committing. Twilio is the broadest telephony platform, covering voice, SMS, WhatsApp, and email with global number coverage, though its pricing model and setup complexity are higher than AgentPhone.
AgentPhone is best for backend engineers and AI agent developers who are writing custom agent code and need phone capabilities without building a full telephony stack. Specific ideal users include developers building AI receptionists, sales agents that follow up leads by text, on-call automation systems, and coding agents that need to receive verification codes. The platform is not suited for non-technical users who need a plug-and-play solution, or enterprise teams that require SOC 2 Type II certification before procurement. International use cases are also not supported yet.
Yes. AgentPhone is API-first: the entire product is accessed via a REST API plus official SDKs for Python and Node.js. There is no significant no-code interface. The platform also ships an MCP server for direct integration with AI coding tools including Claude Code, Cursor, Windsurf, Zed, and any generic MCP client. Developers can provision numbers, configure webhooks, and send outbound SMS entirely via API calls. MCP integration means developers using Claude Code or Cursor can provision phone numbers for their agents directly within their coding environment, without switching to the AgentPhone dashboard.
AgentPhone and Retell AI serve different ends of the telephony-for-AI market. AgentPhone is a bare-metal API that gives agents a phone number and routes calls to a webhook; it requires code to function and has no built-in conversation logic, call analytics, or CRM integrations. Retell AI is a full platform for building and deploying voice agents with a no-code flow builder, latency optimization, call recording, and integrations with Salesforce, HubSpot, and Zapier. AgentPhone is the better pick for engineers writing custom agent code who want minimal abstraction; Retell AI is better when teams need a production-ready voice agent with analytics without building everything from scratch. Retell AI also publishes its pricing ($0.07-0.11 per minute), making it easier to budget compared to AgentPhone.