Web & AI Analyst
This section documents my experiments and learning journey with Web AI technologies, including MCP, agent interfaces, and AI-native content systems.
I built an “article assistant” that can answer questions about a page using local AI in the browser — with a cloud fallback when needed. This post is a conceptual walkthrough of how it works, from on-device inference with Gemini Nano to a simple capability-based routing layer.
Websites may soon expose capabilities directly to AI agents. In this post I experiment with WebMCP on my personal site, implementing two browser-side tools that allow an AI assistant to search an article and subscribe a user to my newsletter.
I built an AI chatbot for my personal website called Ask Ricmac. In this post I explain how it works and how I used Cloudflare Workers, Vectorize, D1, Workers AI and MCP to build it.
As AI agents begin interacting directly with websites and services, a new “Web AI stack” is emerging. Here’s my attempt to map the technologies shaping this next phase of the web.
This week, Google launched A2UI, an open source project to help developers build “agentic user interfaces.” It’s the latest in what’s becoming a regular release cycle of new standards and protocols for building a user interface in AI agents and chatbots. First, there was MCP-UI, a Model Context Protocol (MCP) ecosystem project closely aligned with … Read more
At this month’s Google Web AI summit, the primary focus was client-side AI in the browser — but we also heard a lot about Model Context Protocol (MCP) and its web variant, WebMCP, agents, and how “compute abstractions” like WebGPU and WebNN are enabling web developers to access device hardware. After an introduction by Google’s … Read more
The Google Web AI summit was held earlier this month as an invite-only event in Sunnyvale, Calif. After the event, I caught up with the organizer, Jason Mayes, who leads Web AI initiatives at Google. The last time I’d interviewed Mayes was February 2023, when our main topic of conversation was TensorFlow.js — a JavaScript … Read more
At its annual DevDay event, OpenAI announced several initiatives for developers — who now number 4 million, according to the company. Of most interest was the new Apps SDK, which allows third-party developers to build web-based applications that run as interactive components inside ChatGPT conversations. OpenAI says it will “begin accepting app submissions for review … Read more
This year, MCP (Model Context Protocol) has become the glue that connects AI to the web. Following MCP-UI and NLWeb, we now have another new MCP-related open standard: WebMCP, an initiative of Microsoft and Google. To find out more about WebMCP, I conducted an email interview with Kyle Pflug, group product manager for the web … Read more
It was only a matter of time before Google started turning its market-leading Chrome web browser into an “AI browser.” Today, Google began rolling out Gemini in Chrome, starting with Mac and Windows desktop users in the US. It comes with a bunch of new AI functionality — including AI Mode from the address/search bar, … Read more
In his new memoir, This Is For Everyone, Sir Tim Berners-Lee makes it clear he thinks AI agents are the future of the web. “In the near future, our lives will be transformed by AI ‘agents’ which will interact with the web and take actions to achieve specific goals,” he writes in the introduction. “I’ve … Read more
Earlier this week, we interviewed the two developers of MCP-UI, a new way to add React and web components to AI agents. One of them, Liad Yosef, worked for Shopify at the time MCP-UI launched (just three months ago). It’s no surprise, then, that Shopify did one of the first large-scale implementations of MCP-UI. Today … Read more
MCP-UI is a new open source project that provides a way to add web-based user interface components to AI agents. It builds on the already very popular Model Context Protocol (MCP), which standardizes API access for agents. MCP-UI could become a massive deal for the internet industry. To use professional sports terminology, MCP-UI has a … Read more
Webpages and websites are increasingly seen as “old world UIs” to proponents of AI agents and the Model Context Protocol (MCP). This week, in a presentation at Microsoft’s MCP Dev Days entitled “The Future of User Interaction,” Kent C. Dodd even went as far as to say that “we’re entering a new phase where, instead … Read more
Is RAG dead yet? That was the tongue-in-cheek question posed recently by Douwe Kiela, CEO of Contextual AI. And Kiela knows a thing or two about RAG (Retrieval-Augmented Generation) — he led the team at Meta that introduced the technique in a May 2020 research paper. Of course, Kiela’s post was a reaction to the … Read more