Leadership and governance
Once people are experimenting, the leadership questions start. Who decides what gets scaled? How do good ideas get rewarded when they come from the edges of the organization? How do people build the skills, and what happens to their jobs when the work gets faster? Eventually someone asks whether any of it is paying off.
My bias throughout is to push decisions and learning closer to the work, with a small central team that sets guardrails and builds shared platforms instead of approving everything. Start with where decisions should sit, then governance written by people who build things.
6 posts, in suggested reading order
Push AI decisions closer to the work
Routing every AI decision through a central team creates a queue. The centre should set guardrails and build platforms, and teams should decide.
4 min read · Updated September 2026
AI governance without the theatre
Governance people route around protects nothing. Start with a sandbox, a few data tiers, risk-based review and logging, and write policy from what breaks.
6 min read · Updated September 2026
Reward the idea, not the rank
If a junior analyst and a CEO have the same good idea, they should get the same reward. How I'd set up an innovation fund that works that way.
4 min read · Updated September 2026
Reskilling for AI happens on the job
Courses and central prompt teams can't build AI skills at scale. People learn by using AI on their own work, with a sandbox, peers and a few hours of foundations.
6 min read · Updated September 2026
When AI frees up time, redesign the job
Spare capacity from AI is different from redundancy. Map the work, split it between people and AI, and put the freed time into growth or quality.
5 min read · Updated September 2026
Measuring AI when ROI doesn't fit
Payback-period ROI suits projects that replace a known process. A lot of AI isn't that. Here is what I'd track alongside it, in terms finance can audit.
4 min read · Updated September 2026