Insights
Occasional, considered writing on what actually makes enterprise AI work — and what quietly makes it fail. No hot takes. No hype cycle commentary.
Post-mortems of stalled AI programs converge on a pattern: the technology performed, and the organization around it didn't move. The missing piece isn't compute or talent. It's context — and context has to be built, not prompted.
READ THE PERSPECTIVE →Every enterprise has data. Almost none have encoded the understanding around it. The difference explains most of the gap between AI demos and AI systems.
What changes when agents stop being tools people use and start being colleagues processes rely on — and what leadership structures that requires.
Twelve questions that separate governed deployment from expensive improvisation. Bring them to your next vendor meeting — including ours.
Autonomy is a dial, not a switch. How grounding, confidence scoring, and human-in-the-loop routing turn agents into systems risk teams can sign off on.
Models will keep commoditizing. Organizational intelligence won't. Why the durable AI advantage belongs to the enterprises that encode how they think.
Your most experienced people retire with the playbook in their heads. What it takes to capture judgment — not just documents — before it walks out.
Articles publish on a considered cadence. For advance access, mention it when you get in touch.
Beyond Reading
A discovery conversation applies this thinking to your specific functions, data, and constraints.