Most AI engineers can hand you a tool. Fewer can tell you where it should stop. I do both, because I spent 18 years where getting it wrong had real consequences.
The legal work taught scrutiny. The engineering work turned it into systems people can use.
Law and property work across Scottish legal firms, where detail, records and consequences mattered every day.
Software engineering, automation and, since 2023, AI agent delivery across retail, legal services, charities and startups.
The legal years are why my agents come with boundaries, approval gates and synthetic test coverage, not clever prompts dressed up as a strategy.
For legal services, retail and operations teams where the details matter.
Useful AI needs good engineering and honest boundaries. These are the principles I bring into the room early.
Agents with a clear job, grounded context and outputs people can rely on.
Decision boundaries and approval gates where the stakes are real.
Clear communication, regular updates and technology people understand.
