The race to build the next WordPress

A stubborn giant — and why it matters

It has been reported that WordPress still serves roughly 40% of all Internet traffic. Old? Yes. Clumsy? Often. Ubiquitous? Absolutely. Built before modern cloud plumbing existed, WordPress nonetheless became the layer people touch, fork, and build businesses on — an ecosystem of plugins, themes, and agencies that turned a consumer need (everyone wanted a website) into an industry. Igor Zalutski argues in a recent blog post that the same pattern is reappearing around LLM-based agents: the infrastructure will matter, but the layer users interact with — the “harness” — will make the money and shape the market.

WordPress is the analogy, not Apache

Why compare agents to WordPress and not Apache? Because Apache is a runtime; WordPress sits on top of a runtime and becomes the product everyone customizes. Zalutski (and others) suggest that agent harnesses — tooling that makes agents usable, forkable, and extensible — are the real prize. Sandboxes and microVMs are the modern server; the harness is the thing people will touch, buy, and build businesses on. Allegedly, this is where the next big ecosystem will form: not in low-level isolation tech, but in the platforms that let non-experts own and operate agentic workloads.

The race is on — who wins?

It has been reported that builders in Silicon Valley and beyond are already experimenting with agent harnesses, sandboxes, and runtimes (projects like OpenClaw/Hermes get name-checked as early movers). The question now: will a single platform aggregate plugins and developer mindshare the way WordPress did? It’s early days — think early 2000s web all over again — and that’s thrilling and terrifying. For startups, investors, and developers it’s a classic land-grab moment: build the infrastructure, or build the harness that everyone else runs on. Either way, someone is going to own the developer ecosystem this time around.

Sources: opencomputer.dev, Hacker News