AgentMail raises $6M to build AI email assistants

A few years ago, AI assistants were mostly chatbots that could use basic tools. People were interestbut it was given reliability and security concernsand cost, the technology remained in the early adopters.

How things have changed. Coding providers such as Claude Code, Codex, and Cursor initially saw a lot of appeal, encouraging adoption among programmers around the world, but today we have people using AI to do everything from debugging and building marketing campaigns, managing calendars and organizing meetings. OpenClawThe blockbuster launch earlier this year only made things worse, opening up the door to AI assistants by allowing users to drive only their favorite vehicles at all times.

And if the technology industry we have to believeAI assistants are set to become more and more like real people on the Internet, using apps and services, talking and shopping on your behalf, and often performing many tasks for themselves.

AgentMaila San Francisco-based startup, sees the future moving forward, which is why they created emails designed specifically for AI agents. The company offers an API platform that allows you to provide AI assistants with their inboxes, with the help of two-way conversations, sorting, sorting, labeling, searching, and replying.

The company on Tuesday said it raised $6 million in seed funding led by General Catalyst, with participation from 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).

Along with the investment, AgentMail also announced a login API so you can direct your AI agent to sign up and automatically create an email. The platform also allows you to set up and manage inboxes, permissions, permission scripts, and API keys manually.

According to co-founder and CEO Haakam Aujla (pictured above, right), AgentMail was built from the ground up to give AI agents the same inbox experience that people get with services like Gmail or Outlook — except without the UI tools that humans need. (Note: The platform offers a user-friendly interface, too, for managing different inboxes, and reading and sending emails.)

“When you open Gmail, you have a lot of threads, and within each thread, you can have a lot of messages; those messages can have attachments. You want to be able to write them, search them, filter them, reply, forward,” Aujla told TechCrunch. “We thought we wanted our agents to do that, but they shouldn’t have to, you know, click buttons on the window, because it’s too difficult for agents to do. They should just make API calls.”

Since launching as part of Y Combinator’s Summer 2025 batch, the company has attracted tens of thousands of users, with hundreds of thousands of “support users,” Aujla said, and more than 500 B2B customers.

The early days were slow, however, as the AI ​​assistants had not yet taken off. AgentMail, therefore, focuses on B2B use cases for companies that want to do things like grow their email. But when OpenClaw (then known as Clawdbot) went live in late January, AgentMail saw its user base triple that week, and quadruple in February as people began looking for a way to inbox their emails so they could do more on their own.

The timing was right, as traditional email providers like Gmail put limits on the scope of their email APIs. AgentMail, meanwhile, offers the best free partin addition to paid plans and business subscriptions.

But there’s an obvious problem with handing over email inboxes to AI agents: it makes misuse easy. In order to combat abuse, Aujla said that AgentMail has a limited system: Mailboxes can send 10 emails per day only if it is confirmed by a person; the platform sets value limits if it detects abnormal levels of high activity from inboxes; controls inflation; and creates a sample of new accounts at random to filter keywords.

Aujla says that beyond providing a way to send and receive emails, the main purpose of AgentMail is to work as an identity for AI agents: “We want to give agents the ability to use emails like people do, right? But the idea is, what people use email for is not interactive. It’s your identity (…) let’s just use what already works for people, and what’s already integrated in the rest of the Internet.”

Aujla concluded, “You give the agent an email, (and) now they can use any program that’s already there.”