We’ve raised $22m to scale our browser infrastructure platform for AI agents

I’m excited to share that Kernel has raised $22 million in Seed and Series A funding led by Accel, with participation from Y Combinator, Cintrifuse Capital, Vercel Ventures, Refinery Ventures, and SV Angel.

We’re also thrilled to have the support of incredible angel investors who’ve built some of the most important developer platforms of the past decade: Paul Graham (Y Combinator), David Cramer (Sentry), Solomon Hykes (Docker), Zach Sims (Codecademy), and Charlie Marsh (Astral).

Why we’re building Kernel

LLMs make it possible to automate any work done on a computer, from connecting disparate systems to automating data entry. Yet, cloud infrastructure remains one of the largest challenges preventing developers from unlocking this new technology.

Kernel fills this gap. We provide browsers-as-a-service so AI agents can use the internet the same way people do. Our edge is our focus on performant cloud infrastructure: with Kernel, developers can spin up browsers in milliseconds, persist state across workflows, build agents that securely handle user credentials, and gracefully support human-in-the-loop interactions.

Start-ups and public companies alike already use Kernel to run production workloads, including customers like Cash App, Rye, and countless YC start-ups.

What’s next: enabling sanctioned agents

Our Seed and Series A gives us the resources to accelerate our product roadmap and support our customers’ scaling needs.

As we look to the future, what’s clear is that AI agents need secure and standardized ways to act on behalf of end users. As part of this, I’m excited to announce Kernel Agent Authentication — an identity and permissions layer that lets developers safely authorize agents to take actions on end user accounts, with full auditability and scope control. This makes it possible for AI agents to operate in production environments with the same security guarantees as any other trusted system component. We’ll be announcing partnerships soon as we bring more platforms, identity providers, and developers into the ecosystem.

In addition, we’re already working with customers to provide new tooling that addresses industry-wide challenges, including:

• Enabling Computer Use agents — pushing the industry forward by making Computer Use faster and more reliable, with better observability.
• Expanding Model Context Protocol (MCP) support — expanding support for MCP so that LLMs can access a browser like any other tool call
• Partnering with end websites — giving websites the ability to opt in to, detect, and shape how agents interact with them.

These are all steps toward our broader vision of making Kernel the OS-level infrastructure platform where sanctioned AI agents can securely interact with the web.

If you want to try Kernel, you can get started here. And if you’re excited about building critical infrastructure for AI agents, we’re hiring!

Catherine
Co-founder & CEO of Kernel