AgentMail secures $6M to create an email platform designed for AI agents

Only a few years ago, AI agents were mostly limited to chatbot-style systems that could use simple tools. Interest was there, but concerns around reliability, security, and cost kept the technology largely in the hands of early adopters.

That has changed quickly. Coding agents such as Claude Code, Codex, and Cursor were among the first to gain meaningful traction, driving adoption among developers worldwide. But today, people are using AI agents for a much broader range of tasks, from debugging at scale and building marketing campaigns to managing calendars and scheduling meetings. OpenClaw's breakout debut earlier this year accelerated that shift by making AI agents more accessible and enabling users to run their own localised, personalised agents around the clock.

And if the broader tech industry's predictions hold, AI agents are likely to become as common online as human users, using software and services, communicating and shopping on people's behalf, and automating large portions of digital work.

San Francisco startup AgentMail is betting firmly on that future, which is why it has built an email platform specifically for AI agents. The company offers an API-based service that allows developers to give AI agents their own email inboxes, complete with two-way communication, parsing, threading, labelling, searching, and replying.

On Tuesday, the company said it had raised $6 million in a seed round led by General Catalyst, with participation from Y Combinator, Phosphor Capital, and angel investors including Paul Graham, HubSpot CTO Dharmesh Shah, Supabase CEO Paul Copplestone, and Ramp CTO Karim Atiyeh.

Alongside the funding, AgentMail introduced an onboarding API that can be directed to an AI agent, allowing the agent to sign up and create its own email inbox. The platform also allows users to manually configure and manage inboxes, permissions, allowlists, and API keys.

According to co-founder and CEO Haakam Aujla, AgentMail was built from the ground up to provide AI agents with an inbox experience similar to that of humans using services like Gmail or Outlook; however, it still lacks the visual interface elements people need. The platform still provides a human-friendly interface for managers' inboxes and for reading or sending messages.

"When you open Gmail, you have a bunch of threads, and inside each thread, you can have many messages; those messages can have attachments. You want to be able to label them, search for them, filter them, and reply to "ard," Aujla's id. "We thought we wanted our agents to be able to do that, but shouldn't have to, you know, click buttons on a screen, because that's pretty clunky for agents to do. They should just be able to make API c" lls."

Since launching as part Combinator'stor's Summer 2025 batch, AgentMail has attracted tens of thousands of human users, along with hundreds of thousand" of "agent u "ers," Aujla said. The company also says it now has more than 500 B2B customers.

The early days were slower, however, because AI agents had not yet meaningfully taken off. So AgentMail initially focused on business use cases for companies that wanted to scale things like email communication. But when OpenClaw, then known as Clawdbot, exploded in popularity in late January, AgentMail saw its user numbers triple in that same week and then quadruple again in February, as people started looking for ways to give AI agents email inboxes so they could work more independently.

The timing helped, especially since traditional email providers such as Gmail impose limits on rate and volume through their APIs. AgentMail, by contrast, offers a relatively generous free tier, as well as paid plans and enterprise offerings.

Still, there is an obvious concern with giving AI agents email inboxes: it creates opportunities for abuse. To reduce misuse, Aujla said AgentMail has introduced several safeguards. Agent inboxes are limited to sending 10 emails per day unless a human user authenticates them. The platform also imposes rate limits when it detects unusually high activity, monitors bounce rates, and randomly samples new accounts to look for sensitive keywords.

Aujla said that beyond simply allowing bots to send and receive, AgentMail's broader goal is to serve as an identity layer for "I agents. "We want to give agents the ability to use email in the same way that humans do, right? But the idea is that what humans use email for isn't even communication. It's your identity […] several startups are trying to build new identity protocols for agents. Still, our thesis is to use what already works for humans, and what already is so deeply integrated into the entire internet." Aujla" added, "Yo" give an agent an email address, [and] it can now use essentially any software service that already exists."