Forbes: The Web Wasn’t Built For Machines

Developers are often the first to reveal whether new AI infrastructure works as promised—or collapses under its own ambition.

The internet wasn’t designed for machines. It’s a patchwork of dynamic pages, logins and unpredictable code — chaotic enough to make even the smartest AI stumble.

That gap between intelligence and execution is where the next wave of AI innovation is emerging. Agents can plan and reason, but when they try to interact with the real world, most still fail at something as basic as navigating a website. The web is the unsolved problem of automation.

That’s where Anchor Browser enters the picture. The startup just raised $6 million in a seed round led by Blumberg Capital and Gradient (Google’s AI venture arm) to tackle one of AI’s least glamorous but most urgent challenges: making it safe and reliable for AI agents to use the web.

Anchor isn’t building another chatbot or productivity tool—it’s creating a layer of infrastructure that lets AI operate across real web interfaces with the same precision and predictability as a human.

A Browser Built for AI Agents

When I spoke with Idan Raman, Anchor’s CEO and co-founder, he described the company’s focus as pragmatic rather than flashy. Anchor isn’t chasing novelty — it’s solving a very old problem in a new context.

For decades, the browser has been the center of enterprise work. Most jobs today are essentially a collection of web-based tasks: filling forms, updating dashboards, approving transactions. If AI is going to become a real part of the workforce, Raman said, it needs a way to perform those same actions — safely, repeatedly and within governance boundaries.

That insight led Anchor to reimagine the browser not as a user application but as a cloud-based execution layer for machines. Every AI agent gets its own secure, independent browser environment, hosted in the cloud and isolated from human users.

Rather than letting AI “wing it” with live interactions, Anchor’s system — known as b0.dev—lets agents plan workflows once and execute them reliably, over and over. That shift from improvisation to planning is subtle but powerful. It turns chaotic automation into something closer to enterprise software engineering.

The Browser Wars Return — This Time for AI

After two decades of quiet dominance by browsers like Chrome, Edge (and its sibling/predecessor Internet Explorer) and Safari, the browser wars are back — but for very different reasons.

Perplexity recently launched its Comet browser, and OpenAI followed with Atlas, an AI-powered browser that merges real-time web navigation with GPT’s reasoning. The battle isn’t about faster tabs and slicker design; the real contest is over who owns the doorway between AI and the internet.

Traditional browsers weren’t built for this. They assume a human behind the keyboard, capable of judgment and context. But AI agents don’t think in pixels — they need a consistent, programmable environment with clear boundaries.

Anchor represents a different branch of this evolution — one grown for machines, not people. Its purpose is to give AI a safe, repeatable way to handle the messy real world of the web. As Raman put it, “The web was never built for machines. Someone has to build the bridge if AI is going to join the workforce.”

Read more