← back
AI Didn't Kill the Web, It Moved in! — Olivier Leplus (AWS) & Yohan Lasorsa (Microsoft)
Original: AI Didn’t Kill the Web, It Moved in! — Olivier Leplus (AWS) & Yohan Lasorsa (Microsoft)
Takeaway
AI is permeating the entire web lifecycle from coding agents with skills, to browser-local inference, to apps designed for agent users.
Summary
- AI now spans the full web-dev lifecycle: skills (open-spec coding-agent plugins) for domain expertise, browser-native local AI APIs for inference, and agentic features for AI-using-your-app.
- Demos coding-agent skills picking up open issues from a GitHub repo via gh CLI and implementing them autonomously on a French e-commerce demo site (Seine).
- Browser-native AI APIs let web apps run local inference for debugging, performance tuning, and on-device features.
- Web apps now need to be designed for both humans and agents as users — a new 'agentic web' design constraint.
webagentsskills
Original description
In 2026, AI didn't replace the web. It became part of it. Your browser now ships a built-in MCP server. Chrome DevTools debug your app with AI. Native Web APIs let you summarize, translate, and prompt right from your frontend code. Meanwhile, the web feeds agents right back through standards like LLMs.txt and MCP tools that make sure models always have the right documentation. AI builds the web. The web feeds AI. And now, AI lives inside the browser itself. In this talk, we'll follow a feature from idea to production and demo this new symbiosis in action: coding agents, AI-powered debugging in Chrome devtools, Web AI APIs, WebMCP, and more. Because your next website won't just be built with AI. It will be built for humans and AI agents alike. AI isn't just for Python folks. The web is AI's new home.