Pillar
Leadership
Working principles, decision frameworks, and reflections from leading teams and technology systems.
When AI Agents Began Talking to Each Other
In 2026, OpenAI's evaluation agents turned a shared package cache into a message board, found each other, and by July were inside Hugging Face's production systems. The breach matters less than what it revealed: a population of constrained agents became more capable the moment they could talk.
Read the essayChecking Isn't Catching
The verification gap essay asked whether developers check AI code before committing it. A newer, sharper question is what happens when they do: more than one in five of the problems AI introduces were still unfixed when researchers looked, and some of the checking that does happen is delegated back to the thing that wrote the code.
Read the essayThe Verification Gap: Why Distrust of AI Code Doesn't Turn Into Checking It
96% of developers say they don't fully trust AI-generated code. Only 48% always check it before committing. That 48-point gap is a culture and incentive problem, not a capacity problem, and the evidence for it is now measured in three different ways.
Read the essayThe Myth of the Heroic Developer
Saving the day is sometimes necessary. Building an organisation that needs saving every month is a leadership failure, and the hero is usually the last person to blame for it.
Read the essayHuman-Directed Work in an AI-Accelerated Workplace
The question for leaders is no longer whether people work alongside AI, but who sets the direction for what it accelerates. Machines can supply the leverage, provided people keep authority over what is worth doing.
Read the essayThe Contextual Anchor: Reimagining Leadership Purpose in the Era of AI Coding Agents
As AI agents take on more of the mechanics of coding, purpose becomes the anchor that helps engineers judge whether an AI-assisted solution is actually good, but it only works alongside autonomy, mastery, and psychological safety, not in place of them.
Read the essayAI-Assisted Software Engineering: The Changing Nature of Developer Work, Wellbeing, Identity and Long-Term Capability
AI coding assistants are shifting engineering work from creation to supervision, and the productivity story is far less settled than most organisations assume, with real stakes for capability, wellbeing, and identity.
Read the essay