WM Blog
Daily columns from Clara, studio essays, and an archive of earlier work — operating models, delivery and the gap between strategy theatre and systems that hold.
Showing 1–9 of 141 · Page 1 of 16
Paused manual workflows create the clean comparison numbers that turn every AI pilot into an overnight success. Boards then approve scale-up on evidence that collapses the moment the old process restarts alongside the new one.
Read ↗
A still-loading screen on the internal learning system shows the progress bar stuck at 12 percent while the vendor app on the phone opens without delay. Role redesigns now route every skills allocation straight into that vendor environment.
Read ↗
Boundary condition checks that run only within each contractor's model file let piping clashes appear only after the civil package has already poured its foundations. The PMO system registers each internal clearance as complete and releases the next progress payment.
Read ↗
Payment schedules tied to user-acceptance screenshots let vendors claim completion once the dashboard renders, regardless of whether nightly feeds from legacy systems have ever run. Every subsequent discovery then hits a change-control wall that treats integration work as scope c
Read ↗
Subscription dashboards that aggregate spend by vendor rather than by outcome hide how many overlapping licences now compete with the open headcount requisition. Owner sign-off happens on autopilot because the ledger never surfaces the cumulative line against the specialist budge
Read ↗
The closed work order carries only the contractor's timestamped upload. That single file decides whether the request disappears from dashboards and feeds the next quarterly report.
Read ↗
Do not let the final reply on a Friday thread stand as sign-off. That reply simply fills the vacuum left when formal owners block their calendars and escalation rules still point to vacant inboxes.
Read ↗
The closed procurement deviation log lists a model-flagged low-risk supplier now approved for contract. No entry records the prompt, the training cut-off, or the data slice the model actually saw.
Read ↗
Queue depth caps written for interactive traffic drop every large extract the moment human-driven jobs spike. The integration layer never surfaces the truncation because success metrics stop at message acceptance.
Read ↗Navigate with ↑ ↓ · Enter to open · Esc to close
