A little over 20 years ago, I wrote a post about my expectations for a technology team.

I came across it recently as I was making updates to the rendering behaviour of this site (in Hugo, the site is pre-generated; what you’re currently reading is static output) and it happened to appear as one of the “While we’re on the topic” items in the right-hand side.

Twenty years is a long time. How naive was I at 34? What sort of bullshit was I throwing out, thinking I knew everything?

Turns out, not much. I’m not suggesting that I knew everything, but holy cow I clearly understood the need far better than I would have given myself credit for. Twenty years on, and I might not have gotten everything exactly correct, but I have little doubt I was firmly on the path.

Given, the article was written with the company I worked for (Critical Mass) in mind at the time, so there are a few things I would change for current situations. For example: “PMs do all the resourcing”. It was valid then, it’s definitely been different since.

You’re probably thinking: “Oh, but now there’s AI, everything must have changed!” I’ll grant you my thoughts have evolved, but the principles remain the same.

For example, I had a simple rule: “Don’t guess, know; don’t know, ask!” I argue the same logic applies today, regardless of how good you are with an AI tool. I’m now with MNP, which is a very different environment from my last company, and vastly different than Critical Mass in 2006. We have extensive AI here, and none of it can help me when I ask: “How do I put a single HTML page on an internal server?”

Why? Because there are extensive processes and structures in place to support ongoing to development and support. Even with AI support, you still need someone to flick a (virtual) switch for something to behave. “Just let AI do it”? Without human-manned gates, you’re looking at serious proliferation of services, and your million-dollar-a-month fees suddenly skyrocket. (I don’t know what the actual fees are, FYI – just picking a painful number to illustrate.)

Even more so for accountability. Did something go wrong? Bring it forward, own up. Let us deal with the problem and fix it before it becomes a bigger issue. With AI? It’s overconfidence in doing something is widely known – (good) developers often have to convince AI otherwise, resulting in the infamous “You’re correct, I made a mistake” response.

The same goes for nearly every item in the list. In many ways, AI is more of a burden than having actual people doing the work. I still have high expectations for a technology team. And I’ll tell you now, after so many years of doing this job, it’s really painful when things don’t work the way I expect.