Notes from our work.
Things we've learned, some of them the hard way, written down. Mostly about AI, value and how work actually changes in retail and consumer goods.
Brief · October 2026What a board should ask before funding AI
Three questions every board should ask, including what would make it a no.
Read the piece
Essay · October 2026Absorption capacity: the capacity that isn't budgeted for
Retailers budget money and technology for change, but rarely how much change their people can take on in a season.
Read the piece
Brief · October 2026Accountability does not delegate
What we learned running campaign agents for a national grocer.
Read the piece
Essay · October 2026When agents become operating participants
What changes once software stops answering questions and starts doing the work.
Read the piece
Essay · September 2026A credible AI recommendation has to be able to say stop
Why the most useful answer in an AI business case is sometimes no.
Read the piece
Essay · September 2026The pilot is not the unit of value
Why a long list of AI pilots tells you very little, and what to count instead.
Read the piece
Essay · September 2026The agent is the easy part
What September's wave of personal AI agents should tell enterprise leaders.
Read the pieceWhat we're writing now
Work, Worth, Workforce, Workbench, Watch
The five dimensions that decide whether AI creates lasting value, and why the emphasis shifts after the first deployment.
Decompose the role before you redesign it
Planner and analyst jobs are bundles of tasks with different exposure to AI. Seeing the bundle comes first.
Redeploy before you reskill
Reskilling often trains people for roles that don't exist yet. Moving them to work that already needs doing is faster and more honest.
Price the workflow, not the model
The economics that matter are per workflow, not per model.
Want these when they come out?
We only send something when we have something worth saying.