Writing

A question at the Lancaster AI Symposium, an answer borrowed from Bloom's Taxonomy — and the bigger challenge underneath it.

AI strategy isn’t picking coding tools. GitHub’s recent outage write-up shows why.

Too much data is a real problem. The LLM use case I keep coming back to — turning dashboards into something a non-expert can actually understand.

Vibe coding looks very different depending on where you sit. Co-presenting at the Lancaster AI Symposium with Don Demcsak.

Our 16th edition just wrapped — over 220 people. What I loved about this year was the diversity of thought.

Different business models, different incentives, different answers to the question of where AI fits.

When your AI safety system fails, you don’t just patch it — you build a validation framework.

Nobody became an admin to click through forms. What if the interface was a conversation?

The engineers getting the most out of AI tools aren’t the ones writing better prompts — they’re managing context.

AI tools make it tempting to do more at once. That’s not always a good thing.

I asked Claude Code to add a menu item "after dbt." It created one called "After DBT."

An error alert fired between two meetings. By the time I looked, the system already had an explanation.

Data ingestion is one of the best low-risk, high-leverage places to start with AI.

Good documentation isn’t just for humans anymore — it’s the on-ramp for AI tools.

What I learned bringing AI into the fire service conversation at TSI.

AI coding tools are powerful, but they don’t replace knowing how software works.

Speaking at Penn State about innovation, discovery, and what it means to build something new.