About

Over the course of my career, from research in comparative cognition to building documentation programs and AI-ready information systems, I’ve repeatedly found myself working on the same kind of problem: situations where a system continues operating under assumptions that no longer match reality.

Sometimes those assumptions involve how expertise is developed and where it lives. Sometimes they involve how information is discovered, trusted, and applied. Increasingly, they involve what happens when AI changes the conditions those systems were built around.

The pattern keeps showing up in different domains. In research, it was about what our methods were actually measuring. In technical writing, it was about who documentation was really designed for. In AI systems work, it’s about which assumptions remain valid when generation becomes abundant and judgment doesn’t.

Different contexts. The same underlying move: find what the system was relying on without naming it, and figure out what needs to change when conditions shift.

This site is an attempt to understand those shifts, document recurring patterns, and develop practical frameworks and tools for navigating them.

How I Got Here

I’ve spent my career at the intersection of knowledge systems, organizational design, and how people find and apply information under uncertainty.

My academic background is in comparative and developmental cognition, the study of how different minds represent and reason about the world. That training shaped how I think about assumptions: where they come from, what work they do invisibly, and what happens when they stop holding.

From there I moved into technical communication and developer experience, building documentation programs and information architectures at companies including Foursquare, TUNE, and Chainlink. The most interesting problems were never about writing quality. They were about what the documentation was actually for, who was consuming it, and whether the assumptions baked into its design still matched reality.

At Chainlink, that question became concrete in a new way. As AI agents became primary consumers of developer documentation, the assumptions that had always been load-bearing, that readers would infer, recover from gaps, exercise judgment about what applied, stopped holding. The documentation hadn’t gotten worse. The consumer had changed. Rebuilding the information architecture for that new consumer improved agent-readiness scores from 65% to 92%.

That project crystallized something I’d been working toward for a while: the interesting problems aren’t documentation problems or AI problems or expertise problems. They’re problems that emerge when systems continue operating under assumptions that no longer hold, and the most important ones are the assumptions nobody noticed because they’d always worked.

The essays and frameworks on this site grew out of that observation. The essays explore the patterns. The frameworks explain the mechanisms. The emerging tools focus on what to do next.

If these ideas intersect with work you’re doing, I’d be glad to hear from you.