The Web Now Has Two Audiences — And One Browser
For decades, the browser has been a human tool. Tabs, bookmarks, themes, extensions — every feature optimised for the person sitting in front of the screen. AI software is rapidly evolving from chatbots that answer questions to agents that complete tasks on users' behalf, and browsers are a critical part of that transition: agents will need to navigate the web and interact with websites much like humans do.
Kitesurf formalizes a division that has been building quietly for years — the web now has two separate audiences, humans and machines, and their requirements have diverged to the point where building for both in a single product no longer makes sense. Cloudflare's answer is to stop trying.
What Kitesurf Actually Is
Kitesurf began as a proof of concept that ported Obscura, an open-source Rust headless browser engine for AI automation, to Cloudflare Workers. It has since evolved into a browser that combines a modular rendering engine from Blitz, Firefox's Stylo CSS parser, and the Boa JavaScript engine — all three compiled to WebAssembly and running inside Cloudflare's V8 isolates.
Cloudflare says it decided to build Kitesurf just 12 weeks ago, and it runs entirely on top of the company's serverless platform, Workers. That speed of development is notable — and a signal of how seriously Cloudflare is treating the agentic workload opportunity.
Kitesurf is available for free while in beta in Browser Run, which lets developers programmatically control and interact with headless browser instances on Cloudflare's network.
The Performance Trade-off, In Numbers
The headline efficiency claim is specific and independently verifiable. In Cloudflare's own tests, Kitesurf used 3.1 to 3.8× less CPU and 4.7 to 7× less memory than Chromium, depending on the task — but took 1.7 to 1.8× longer to complete those tasks.
The trade-off is intentional: for agent workloads like scraping, form filling, and content extraction, the resource savings outweigh the speed penalty. In other words, Kitesurf isn't trying to be fast — it's trying to be cheap at scale.
Unlike traditional web browsers built for humans, a browser built for AI agents doesn't care about visual elements like themes, tabs, or browser extensions. Instead, it needs to manage context windows, performance, token costs, and scalability.
Once a task completes, nothing persists. A human browser keeps your history, passwords, and preferences. Kitesurf keeps nothing, because an AI agent has no use for any of that between sessions.
Compatibility and Security
One major concern with purpose-built runtimes is fragmentation — developers don't want to throw away their existing tooling. Cloudflare has addressed this directly.
Kitesurf speaks the Chrome DevTools Protocol (CDP), making it compatible with Puppeteer, Playwright, and other browser automation tools without requiring developers to change their existing workflows.
It also passes 215,000+ Web Platform Tests despite being a ground-up rewrite in Rust rather than a fork of Chromium.
On the security side, Kitesurf has to deal with a different security model than consumer browsers, including risks like prompt injection, where malicious instructions hidden inside web content could manipulate an AI system.
The threat model in the context of AI using a browser is fundamentally different — new problems like prompt injection and tool safety are top priorities.
Why This Matters for Developers Building Agents
The core pitch is straightforward: if agents are going to navigate websites, fill out forms, extract information, and complete browser-based tasks, developers shouldn't have to build and scale their own browser stack to make that happen.
CPU and memory are among the key cloud resources that drive up the cloud bill — and for teams running hundreds of concurrent agent sessions, the savings compound fast. The free beta period gives developers a low-risk on-ramp to test Kitesurf against their own agent workloads before committing to it as infrastructure.
It's also worth noting the broader competitive context. The desire for a specialised browser for AI agents didn't begin with Cloudflare — in April 2025, OpenAI signalled interest in acquiring Google Chrome, if Google were forced to sell due to regulatory pressure, with the intention of building on it to develop their own AI browser. Cloudflare's move is a different bet: rather than repurposing a human browser, build something native to the agent paradigm from scratch.
Bot traffic now generates the majority of all web requests, with AI agents as the fastest-growing segment, per Cloudflare's own network data from last year. Kitesurf is Cloudflare's bid to own the infrastructure layer of that shift — turning the browser itself from a destination into a utility.
