Background
About
Twenty-five years of building for the web, most of the last decade spent getting technology delivered inside large regulated organisations.
I started building for the web in 1999, which means I learned the platform in the order it was invented: markup, then servers, then databases, then everything layered on since. That order turns out to matter. A lot of debugging is just knowing which layer to suspect.
The bulk of my engineering career has been full-stack work on the LAMP side – PHP, MySQL, Linux, Apache – alongside Python, .NET and C#, and React Native for mobile. I have led development teams, built cross-platform native applications for iOS and Android, and spent a long time on data-heavy sites where the interesting problems were in the database rather than the interface.
The move into program management
Somewhere along the way the constraint stopped being my own throughput and started being everything around it: approvals, dependencies, vendors, and the gap between what a steering committee was told and what was actually true. So I moved toward the part of the work that decides whether software ships at all.
That has meant technical program and project management for large regulated businesses, including Fortune 50 banks and investment firms, at Trellist Marketing and Technology. Financial services is an unforgiving environment to deliver in and a useful one to learn in – the controls are real, the scrutiny is real, and nothing gets waved through because the demo was good.
Why AI, and why now
I have been through enough hype cycles to be suspicious of this one. What changed my mind was not a demo but the ordinary experience of using these tools on real work: a legacy codebase documented in an afternoon instead of a fortnight, test coverage built retroactively for a system that had none, a full application delivered in the time a proposal used to take.
None of that is magic, and none of it removes the need for someone accountable who can read the output. What it does is change the arithmetic on a large class of work that was previously too expensive to justify. Most organisations have not repriced that work yet.