API-driven phone outreach for AI agents

AIThis description was written by a model from public signals. The facts behind it are listed under Evidence.

Problem

AI assistants and agents cannot reliably reach businesses or access real-world information because most communication still happens over phone calls, which require navigating menus, waiting on hold, and handling voice interactions. This limits what agents can accomplish in the real world.

Solution concept

A ready-to-use API that agents and apps can call to perform phone-based tasks: given a plain English task and phone number(s), the API navigates voice menus, waits on hold, extracts structured information, and returns outcomes with recordings and transcripts. It should handle parallel calling and work out of the box without custom voice agent development.

Who has it

Developers building AI agents, assistant applications, and companies building products that need to interact with local businesses or service providers via phone calls.

How they handle it today

Currently, when an AI agent or app needs to contact a business, developers either build custom voice agent infrastructure, rely on manual workarounds, or accept that the agent cannot complete the task. Some existing tools (like Muse with beta call support) can only call one place at a time sequentially, and general voice agent builders require significant custom setup work.

Why now

AI agents and assistant applications are becoming mainstream tools that users expect to handle real-world tasks, but they are bottlenecked by the phone as a communication interface. The founding team identified this gap through building their own agent product at Google Maps and seeing the same problem across multiple use cases.

How it could charge

Likely usage-based pricing per API call or per phone call completed; the signal mentions 250 free calls as an initial offer, suggesting a metered model.