Topic
Notes / Essays
Shorter notes and working thoughts that don't fit cleanly into any of the other pillars. The point in each is usually a single observation. Some grow into longer memos later. Some stay where they landed. The stakes here are lower than in a pillar piece, and a few of the better ideas on the site started in this pile.
All memos
-
AI made OKRs that lean on output easier to hit and less useful
When output gets cheap, key results start pressuring the part of the work that got easy.
-
The ‘AI replaces juniors’ story points at the wrong layer
The first role under pressure is often the manager who translated senior asks into team work.
-
A weekly structure for a team with uneven AI use
Both halves should leave with something they couldn't have gotten from a model.
-
Coordination tax
What companies pay to keep people in sync, even when the sync doesn't happen.
-
Most ‘AI readiness’ work is information reachability work
The block for most companies is that their own information isn't reachable, even by themselves.
-
The Arabic operators I know still default to English tools
English tools integrate better with the surrounding work, which is already in English.
-
AI programs in the region are running off the procurement template
Three months in, the measure of success is still seat activation, not workflow change.
-
I don’t know how to measure a team where AI use is uneven
Output goes up while learning may flatten. I'm not sure which signal a manager should be tracking.
-
Four questions I ask before a team buys an AI tool
Workflow questions. The tool comes later, if at all.
-
Operator debt
The backlog of decisions an operator hasn't written down yet, compounding quietly.
-
Most AI pilots in the region stall before anyone tests the model
Most of the failure happens upstream, in data access and operator buy, before model quality is even the question.