In 1993, a physicist at CERN released a specification that almost no one read and almost everyone eventually used. HTML was not a programming language. It was not even particularly elegant. It was a way of saying: here is some content, here is how it relates to other content, and here is roughly how it should look.
What made HTML powerful was not the technology itself. It was what the technology assumed. It enabled any machine anywhere running any operating system to parse documents and render them usefully. This was a universal agreement about exchanging meaning. Before HTML, software was compiled, siloed, and closed. After HTML, everything became composable, linkable, and readable by anyone with a browser.
The current inflection point involves a new primitive at a higher layer. The primitive is not a markup language but natural language itself, mediated by a model that parses intent and returns action. The interface layer is collapsing into the instruction layer. HTML gave structure to information. AI gives agency to interfaces. That is a bigger jump than it sounds.
A new runtime for a new primitive
The architectural parallels between the HTML era and the AI era are not superficial. Each layer of the web stack has a direct analogue in the AI stack — and the mapping reveals exactly which layer of the AI stack is still missing.
HTML did not create one wave of disruption. It created six, each building on the last, each birthing industries that did not previously exist. We are, right now, inside the opening of the same structure — and the industry mostly does not know which wave it is in.
HTML did not create one wave. It created six.
Place the HTML timeline next to the AI timeline and the phases align with unusual precision. Each wave of the web had a defining product, a defining craft, and a defining failure mode. The AI era is running the same sequence, offset by about three decades.
The industry currently sits somewhere between wave one and wave two. The chatbot-as-pamphlet era is already feeling tired. The dynamic, grounded, personalized wave is beginning to break. Waves three through six exist as demos and roadmaps. The people building them are doing what front-end developers were doing in 1997: inventing a profession without a name for it yet.
Not a new tool. A new contract.
What separates this moment from innovations like smartphones or cloud computing is that those were delivery mechanism changes. HTML and AI introduced entirely new ways for humans to interact with computers.
HTML gave us the visual interface as the universal default. Before the web, interacting with software meant CLIs, proprietary GUIs, application-specific paradigms. After the web, the page became the contract — the agreed-upon unit of interaction between any human and any software.
The transition feels disorienting because skills do not transfer cleanly. Understanding visual design does not teach graceful conversation design. Knowing CSS does not teach system prompt writing. The craft is genuinely new. We are not making better web pages. We are replacing the web page as the primary unit of interaction.
Every wave creates professions that could not be named before it.
HTML created the webmaster, the front-end developer, the UX designer, the SEO specialist, the growth hacker, the React engineer. None of those titles existed in 1993. Most could not have been described in 1993, because the problems they solve had not yet become legible.
The AI era is creating its own list: the AI product lead who owns the model's behavior end to end; the prompt engineer who turns intent into model behavior; the conversation designer who architects user flows through dialogue; the LLM optimization specialist who makes brands visible inside AI answers; the agent orchestrator who designs multi-step autonomous workflows; the eval engineer who structures feedback loops for model reliability.
Most of these roles barely have agreed-upon names yet. Some are being pioneered by people who do not know they are inventing a profession — exactly where front-end development was in 1997.
AI has no W3C. That is the risk nobody is pricing.
HTML had the World Wide Web Consortium. A body of consensus, slowness, and genuine interoperability standards that made the web buildable. You could write HTML once and trust it would render recognizably across machines, operating systems, and decades. The spec was public. The process was open. There was always a standard to converge toward.
The consequences are already visible. A prompt that works reliably on one model breaks silently on another. An agent built for one API version stops working when the model updates without notice. Behavior that is safe on one deployment is unsafe on the next. There is no "View Source" for alignment. You cannot audit a model like a webpage.
The web's interoperability was hard-won through years of standardization work that felt bureaucratic precisely because it was important. The AI industry is currently racing to ship while that work goes undone. The companies that win will not be the ones that prompt the best. They will be the ones that make prompting irrelevant, and trust portable.
Build the browser, not just the page.
The most durable opportunity in AI is not building the next application. It is building the infrastructure that makes the next thousand applications trustworthy: evaluation frameworks, behavioral contracts, interoperability layers, audit tools. The picks and shovels of an intelligence gold rush.
We are unmistakably at the beginning of something. The primitive has arrived. The waves are forming. The jobs are being invented. The interface paradigm is shifting beneath our feet. What we do not yet have is the governance layer that will make this sustainable — the missing W3C that will let the next six waves build on each other rather than collapse into each other.
Whoever builds that will matter as much as whoever built the browser.