Just a few years ago, most AI agents were chatbots with basic tools. People were interested, but concerns about reliability, security, and cost kept the technology in the realm of early adopters.
How has the situation changed? While coding agents like Claude Code, Codex, and Cursor initially gained the most attention and drove adoption among programmers around the world, people are now using AI agents to do everything from debugging at scale and building marketing campaigns to managing calendars and scheduling meetings. The blockbuster debut of OpenClaw earlier this year accelerated things even further by opening up access to AI agents by allowing users to run their own localized, personalized agents 24 hours a day.
And if the tech industry is to be believed, AI agents will become as common as real humans on the internet, using software and services to talk and shop on your behalf, and generally automate a wide range of tasks.
AgentMail, a San Francisco startup, sees its future firmly unfolding, which is why it built an email service designed specifically for AI agents. The company provides an API platform that allows AI agents to have their own email inboxes, supporting two-way conversations, parsing, threading, labeling, search, and replies.
The company announced Tuesday that it has raised $6 million in a seed funding round led by General Catalyst. Y Combinator, Phosphor Capital, and angel investors Paul Graham, Dharmesh Shah (CTO of HubSpot), Paul Copplestone (CEO of Supabase), and Karim Atiyeh (CTO of Ramp) participated.
Alongside the funding, AgentMail also announced an onboarding API that allows you to point to AI agents so you can directly sign up and create your own email inbox. The platform also allows you to manually configure and manage your inbox, permissions, allowlists, and API keys.
According to co-founder and CEO Haakam Aujla (pictured above, far right), AgentMail was built from the ground up to give AI agents an inbox experience similar to what they get with services like Gmail and Outlook, except for the UI elements that humans need. (Note: The platform also provides a fully human-usable interface for managing various agent inboxes and reading and sending emails.)
“When you open Gmail, you have a lot of threads, and within each thread there can be many messages, and those messages can have attachments. You want to be able to label them, search for them, filter them, reply to them, forward them, etc.,” Aujla told TechCrunch. “We wanted the agent to be able to do that, but not have to click a button on the screen, because that would be a huge pain for the agent. They just need to be able to make the API call.”
Since launching as part of Y Combinator’s Summer 2025 batch, the company has attracted tens of thousands of human users, hundreds of thousands of “agent users,” and more than 500 B2B customers, Aujla said.
However, the early days were slow as AI agents had not yet really taken off. Therefore, AgentMail focused on B2B use cases for businesses that want to expand their email communications and more. But when OpenClaw (then known as Clawdbot) arrived in late January, AgentMail’s user count tripled that week and quadrupled in February as people started looking for ways to give agents an email inbox and let them do more.
The timing was right, as traditional email providers like Gmail had rate and volume limits in their email APIs. AgentMail, on the other hand, offers a fairly generous free tier in addition to paid plans and enterprise subscriptions.
But there are obvious problems with giving an AI agent an email inbox. That means it’s easier to exploit. Aujla said AgentMail has several systems in place to combat fraud. An agent’s inbox can only send 10 emails per day unless authenticated by a person. The platform imposes rate limits if it detects unusually high levels of activity from your inbox. Monitor your bounce rate. We also randomly sample new accounts to filter sensitive keywords.
Aujla says that beyond providing a way for bots to send and receive emails, the larger purpose of AgentMail is to serve as an identity layer for AI agents: “You want to give your agents the ability to use email the same way humans do, right? But the idea is that humans don’t even use email for communication; it’s your identity (…) There are some startups trying to build protocols, but our argument is, let’s use what already works for humans, and what’s already deeply embedded throughout the internet.”
Aujla concluded, “Giving your agents an email address essentially allows them to use all of your existing software services.”
