For Your Information
An experiment: I'm sharing my personal knowledge graph — a map of what I'm reading, thinking about, and connecting across time — so you and your agent can use it to stay close to what's happening in AI. It's called FYI.
People ask me often which sites they should follow to stay current on AI, and I'm always glad to answer. But I've never felt like my answer was quite right, because the sites aren't really the point.
What makes AI hard to follow isn't just the volume, it's the shape of it. Research, products, policy, economics, and culture are all moving at once. Those threads aren't parallel: each changes how the others should be understood, and together they reach into almost every industry. To really follow AI, you need to make connections across all of them, reading for relationships rather than just events.
I read about AI constantly. It's the center of my work. Over time, I noticed something: the most useful thing I was doing wasn't cataloguing everything, it was filtering. Deciding what felt important, what felt surprising, what felt like it connected to something I'd seen three months ago. There's someone in your life whose read on a situation you'd trust over a more complete briefing, not because they know more facts but because their judgment is sharp in a way that's hard to explain. That judgment is partly tacit knowledge: understanding built through experience. You can't simply write it all down, but you can show more of what it produces: the things you notice, the connections you make, and the evidence that changes your mind.
That's what I'm trying to do with FYI: a curated, connected record of what I'm actually paying attention to and why. A feed of someone's attention is often more interesting than a feed of their finished thoughts. When I link two things together, or flag something as challenging an idea I'd previously found convincing, that relationship is part of the record too. What you get here is a specific perspective on some things, with the reasoning visible.
The simplest way to use it is to read it the way you'd read a blog — following along and dipping in when something catches your eye. But you can also connect it to your AI agent, and that opens up something different. Your agent can combine my feed with your own notes and context to surface connections you wouldn't otherwise have reason to look for: something I flagged last week that relates to what you're working on now, shifts in how I read a topic over time, or answers to questions like "has anyone in my feeds changed their mind about agents recently, and why?"
The agent part only works if your agent can actually read the site, so FYI is built for agents as deliberately as it is for people. Everything is public and machine-readable, with no account or API key required. There’s a structured JSON feed, an llms.txt file, and a small query API over the graph itself: search it, pull what’s new, or follow the connections around any idea. Give your agent the instructions at mattwood.fyi/agents and it will know how to read, search, and follow the graph.
You can also set feeds like this alongside each other, comparing perspectives across people the way you might compare notes with colleagues after reading the same article. And that's the idea I'm most excited about. My perspective here is just one view, and it becomes more useful sitting next to yours, or next to someone else whose read on AI you've always found sharp. Each feed is personal, each one gains from connection, and together they can help us see things we'd miss alone.
That's what I'm hoping to find out: whether connected perspectives can help people think better about a fast-moving field. The form will almost certainly change as I learn what people and their agents actually find useful. That is part of the experiment too.
Fire it up, at mattwood.fyi.
PS: The blogs of the early internet were doing something like this long before "knowledge graph" was a phrase anyone used. Microblogs from Swiss Miss, Daring Fireball, Marcel Molina, Jason Kottke, and others were onto exactly this twenty years ago: short, associative, personal feeds of attention rather than conclusions. That spirit is very much what FYI is trying to be, for the agentic age.