Oct 13, 2026
•
6 min read
Why customer-specific AI work turns into a services business unless the product has a defined place to put it.
Oct 6, 2026
Why the usual incident investigation methods fail for AI features in software products, and what to record instead.
Sep 29, 2026
7 min read
Why safe AI agent access often requires moving business rules out of the screen.
Sep 22, 2026
AI can cut the cost of writing code. Some estimates assume the rest of the program got cheaper too.
Sep 15, 2026
8 min read
The Revenue Substitution Score rates each revenue line on three dimensions the general ledger does not report, and the band it produces sets what the operating plan may assume about that line.
Sep 8, 2026
Running the same agent on the same task can consume up to thirty times more tokens on one run than another. The Authorship Split separates the spending somebody inside the company decided from the spending nobody there did, because only one of the two can be approved.
Sep 1, 2026
5 min read
The Repricing Gap shows which contracts lose margin before the company can act.
Aug 25, 2026
AI investment serves three different economic objectives. Boards should see the allocation across all three.
Aug 18, 2026
The Attribution Tree sorts every AI line in the FY27 plan into one of five verdicts: park, instrument, prove, stop, or scale.
