WEB TECHNOLOGY
The web was built for humans to navigate with browsers. NLWeb is an attempt to rebuild it for AI to navigate with natural language.
The concept is straightforward: instead of websites presenting information in HTML designed for human eyes and search engine crawlers, they expose a structured, semantically rich interface that AI agents can query directly in plain language. “What are your opening hours on bank holidays?” becomes a question the website can answer directly to an AI assistant, rather than requiring the AI to scrape and interpret a page designed for humans.
Why This Matters
The current relationship between AI assistants and web content is awkward. AI systems either use pre-scraped training data (which is static and quickly outdated), call search APIs and interpret the results (which is slow and lossy), or scrape pages directly (which is fragile and often blocked). NLWeb, if widely adopted, offers a cleaner alternative: a standardised interface that websites maintain themselves.
“NLWeb is essentially a new layer of the web — one built for machines to query rather than humans to browse.”
The Adoption Challenge
The technical specification is interesting. The adoption challenge is harder. The web is built on the principle that anyone can put up a page with minimal technical requirements. NLWeb requires site operators to implement and maintain an additional interface layer — and the incentive to do so depends entirely on whether AI-mediated traffic becomes important enough to justify the investment.
Given the trajectory of AI assistant usage, that point may arrive faster than sceptics expect. But “may” is doing a lot of work in that sentence, and the history of web standards is full of technically sound proposals that never reached critical mass.
Tags: Artificial Intelligence • Opinion • Technology & Society • 192.168.1.22/