Architects were ready for AI before many engineers for good reasons.

𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘀 𝘄𝗲𝗿𝗲 𝗿𝗲𝗮𝗱𝘆 𝗳𝗼𝗿 𝗔𝗜 𝗯𝗲𝗳𝗼𝗿𝗲 𝗺𝗮𝗻𝘆 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 𝗳𝗼𝗿 𝗴𝗼𝗼𝗱 𝗿𝗲𝗮𝘀𝗼𝗻𝘀.

More than two years ago, I started looking at AI seriously with my team of architects to generate code.

At the time, many engineers around us were wary of it and struggled to see the value. To a lot of people, AI still looked like fancy autocomplete.

My architects didn't see it that way. There was little doubt in our minds where this was going.

Even today, I keep seeing articles about how engineers need to change the way they work with AI: think beyond the code, rethink code review, pay more attention to the wider system, and get better at directing rather than simply implementing.

Architects have already had to make much of that transition.

Most architects I know started as engineers. But the role forces you to zoom out:

🔹 How should the system be structured? 🔹 Does it solve the business problem? 🔹 How will it evolve, and where are we creating technical debt? 🔹 What needs to be done now, and what does not? 🔹 When is an imperfect solution the right one because it limits scope and blast radius?

That last one has caused plenty of interesting PR discussions over the years.

An engineering review can produce a list of things that should be fixed because they break a standard or could be more elegant.

The architectural question is often:

𝗬𝗲𝘀, 𝗯𝘂𝘁 𝗱𝗼𝗲𝘀 𝗳𝗶𝘅𝗶𝗻𝗴 𝘁𝗵𝗮𝘁 𝗵𝗲𝗹𝗽 𝘂𝘀 𝗮𝗰𝗵𝗶𝗲𝘃𝗲 𝘁𝗵𝗲 𝗴𝗼𝗮𝗹?

Sometimes the right answer isn't perfect code. It's the best code for the goal, the risk and where you are in the journey.

That translates extremely well to AI.

But there's another architectural skill that matters even more:

𝗕𝗲𝗶𝗻𝗴 𝗮𝗯𝗹𝗲 𝘁𝗼 𝗱𝗲𝘀𝗰𝗿𝗶𝗯𝗲 𝘄𝗵𝗮𝘁 𝘆𝗼𝘂 𝘄𝗮𝗻𝘁.

Knowing what “good” looks like is only half the battle. An architect has to describe it clearly enough that someone else can build it.

🎯 The intent. The constraints. The trade-offs. What matters, and what doesn't. How you'll know when the result is right.

I've mentored engineers moving into architecture, and this is often one of the hardest jumps.

I'd regularly find myself saying:

𝗗𝗼𝗻'𝘁 𝗴𝗼 𝗮𝗻𝗱 𝗱𝗼 𝗶𝘁 𝘆𝗼𝘂𝗿𝘀𝗲𝗹𝗳.

If the team can't deliver what you intended, first ask whether you described it well enough.

Your job is no longer just to apply your knowledge directly. It's to scale your knowledge and experience while someone else writes the code.

Now that someone can be AI.

That's why I think architects were unusually well-positioned for this shift.

🧭 AI rewards many of the same skills architecture already demands: understanding the whole system, knowing what good looks like, making pragmatic trade-offs, providing context, explaining intent and reviewing outcomes.

For years, architects have been learning how to scale themselves through engineering teams.

AI gives us another  extraordinarily powerful way to do it.