AgentPhone

Product & Competitive Intelligence

Gives AI agents real phone numbers for calls and texts.

Company Overview

AgentPhone is a dev tool that gives AI agents phone numbers for SMS, calls, transcripts, and webhooks. Serving buyers across developer tools (agent builders), support automation (AI receptionists), and sales ops (lead follow-up teams).

Latest Intel

Zeitgeist tracks private signals to determine where the company is heading strategically.

View All The Latest Signals

What They're Building

The company's public product roadmap & what they're committed to building.

Agent Phone Numbers

US and Canadian numbers for agents that need a durable identity for calls, SMS, and inbound handling.

Unified Webhook

One event layer routes messages, calls, transcripts, and status changes into a customer backend.

Hosted Voice Agents

Built-in STT, LLM prompting, and TTS let teams run voice agents without stitching together the stack themselves.

MCP Server

MCP support lets Claude Code, Cursor, Windsurf, and other agent clients place calls and send texts from tool calls.

SDKs and API

Python, TypeScript, and REST surfaces make the product feel like agent infrastructure, not a telecom console.

Competitors

Twilio:

Telecom API incumbent with massive carrier reach, but AgentPhone competes by packaging phone workflows for AI agents.

Vapi:

Voice agent infrastructure player focused on real-time calls, while AgentPhone widens the surface to numbers, SMS, webhooks, and MCP.

Retell AI:

Voice AI platform for phone agents, stronger on call automation while AgentPhone pitches a lower-level agent phone layer.

AgentPhone

's Moat:

No credible moat yet, likely path is workflow switching costs as teams wire numbers, webhooks, transcripts, and MCP tools into agents.

How They're Leveraging AI

NLP Context Management

Conversation threading and transcript storage can give agents memory across calls and texts without requiring customers to build the context layer.

Agentic Workflow Automation

Agent webhooks let customers plug their own LLM, RAG, or business logic into calls and texts while AgentPhone handles telecom state.

Speech Recognition

Hosted voice agents use streaming speech recognition, prompt-driven LLM calls, and voice synthesis to let agents hold phone conversations.

AI Use Overview:

AgentPhone appears to orchestrate streaming STT, hosted LLM prompts, TTS, webhooks, and MCP tools rather than training a proprietary model.