Modular journalism meets answer systems. A resilient linguistic framework for news, mapped user information needs, and a minimal atom schema for AI readers. Coming up with an atom-based infrastructure is the easy part. The hard part is getting publishers to understand the structure imperative — rebuilding their CMSs around atoms — and agents to respect structured attribution.
Detecting user effects and mapping Information needs in automated mews pipelines
When AI annotations read better than the original reporting, perhaps it is time to rethink how we consume news. I, for one, would keep my NYT subscription if it were delivered in an AI-mediated format
A ‘geometric proof’ that journalism survives the AI apocalypse if and only if it adopts modular production
Copy and paste the recipe of one of my custom GPT agents, the “pipeline architect.” It will tell you something you may not like to hear: that AI agents do not replace structure and automation; they depend on them, including the very boring plumbing.
AI answers are becoming the default interface to information. In three ordinary searches, I got what I needed from Gemini and almost never reached journalism. That gap is the Gemini barrier: when answers satisfy intent before reporting is discovered. The question is not whether journalists should write, but how they should publish so that answers still lead to them.
I followed a swatch (green, not white) through Gmail's dark mode and tested four color-mapping methods to see what sticks
Engineering had nothing to do with it, and it was about journalism all along.
Journalism & Engineering is a platform for journalists with product and data fluency