Skip to content
← All notes

My Principles for Leading AI-Assisted Development

ai leadership

These principles ensure that as AI takes on more of the “how”, I remain firmly in command of the “what” and “why”:

  • Own the architecture: I define the structure, patterns, and boundaries. AI works within the system I design, not the other way around.
  • Make the hard decisions: Trade-offs, priorities, and judgement calls are mine. AI can inform them, but it can’t weigh what matters to my team and users.
  • Maintain the context: The full picture lives with me: history, constraints, politics, roadmap. Agents only see what I show them, so making context easy to share is part of the job.
  • Understand the why: Systems I can’t explain don’t get shipped. If I don’t understand it, I can’t debug it, extend it, or defend it.
  • Preserve the rationale: Decisions and their reasoning get written down. Institutional knowledge shouldn’t live only in session history.
  • Take responsibility: When production fails at 3am, I’m the one they call. AI doesn’t have a phone.
  • Verify the output: Everything AI produces gets scrutiny proportional to the stakes. I recognise the appropriate level of rigour and calibrate accordingly.
  • Know when to intervene: Recognising when the AI is heading down the wrong path is a skill in itself. Early course-correction saves pain.
  • Maintain your craft: Technical skills need deliberate maintenance. Tools and frameworks change constantly, but the foundations, systems thinking, architecture, and sound engineering judgement, are what everything else in this list stands upon.

I remain accountable, AI is a powerful collaborator, but I’m delegating execution, not leadership. My value is my judgement, and I protect it.

← All notes