Topic
ai
Start with the work, not the org chart
Most companies aim AI at their functions and end up with the old company, accelerated. The harder move is to start from the work itself, and let the org chart come back into the room last.
You cannot lead from the surface
AI keeps moving the boundaries between what teams build, buy and automate. Leaders need enough firsthand understanding to see when they've moved.
The work should leave something behind
AI does not just make people faster. It lets a team convert what it learns into capacity, so that every migration, incident and analysis makes the next one cheaper. A revisit of the four states of engineering teams for an era when capacity no longer has to be linear.
Leadership exists to own what has no natural owner
Teams optimise locally; leaders own the seams between them. AI has raised the price of leaders who will not go deep enough to keep the system coherent.
The tools are democratic. The outcomes will not be.
AI is democratising access, but curiosity, judgement and the willingness to experiment will determine who compounds with it.
Not all AI agents are the same. Confusing them is quietly stalling enterprise AI.
Personal assistants, builder tools and enterprise agents may share the same technology, but they carry different accountability and require different governance.
Beyond AI anxiety: rethink, don't sprinkle
The answer to AI anxiety is not to buy faster, but to understand the work deeply enough to rebuild it from first principles.
AI raises the price of judgement
As AI makes software easier to produce, advantage shifts to judgement: knowing what to build, buy, automate and own.