Aug 18, 2026
•
4 min read
We could check the answer. We couldn't check the reasoning.
Field Notes
Aug 11, 2026
3 min read
Leadership
Aug 4, 2026
2 min read
Issue #10: What a fast kill taught us about where AI actually earns its keep in B2B commerce
Enterprise Apps & AI
Jul 28, 2026
Distributors and manufacturers are posting hard eCommerce numbers this year. The figure everyone quotes is the tip; the durable story is underneath it.
Industry Intel
Agentic AI doesn't answer. It acts. Here's the rollout that earns it the right to — one rung at a time.
The metrics that make an AI look good in a demo are almost never the ones that tell you it's earning its keep in production.
Frameworks
Most failed implementations were diagnosable months before go-live. The warning signs are boring, repeatable, and easy to miss when you're heads-down shipping.
They cost about the same to build. One pays for itself in a quarter; the other becomes a line item nobody will defend. The difference isn't the model.
Issues #1–3 covered the decisions leaders make. This one is about the people who absorb them - why the loudest resister is often right, and why one kind of pushback isn't information at all.
At a recent industry roundtable, the AI questions practitioners actually ask landed in three areas — and the honest answers are not always the ones that are led with.
Issue #1 named the two disciplines mid-market skips. This is the operating manual for the second one — without the vendor-versus-customer framing.
Why your platform replacement earns less than the one it replaced — and the two artifacts that prevent it.
Weekly field notes from the practitioner side of manufacturing IT — Enterprise Apps, AI, and the messy middle between the plant floor and the cloud.