The architects. Sigh. Smile.

“I have some good architect friends, but the vast majority” were real pain in the asses during the golden age when the word agility had a good reputation.

They were the ones who wanted to plan everything upfront (the deadly big up front design), the most Cartesian of Cartesians, the hair-splitters down to 512kb, those who wanted to decide everything without ever getting their hands dirty, comfortable in their ivory towers. Mostly the anti-agile folks.

Yes, I’m caricaturing – just a little – the vast majority, not all. And those who broke free from these flaws had even more value in my eyes. We always loved the architects who got their hands dirty.

But many at the time (2005-2020) were obstacles. Has-beens. Up until today.

Today many continue to go astray (in my eyes), but I look at them with tenderness, because today, they’re not wrong on substance, but on timing—basically they’re defending agile when no one defends it or talks about it anymore.

It’s an open secret, agile retains all its relevance and strength today. Only it has been so poorly understood, so poorly used, so poorly taught (the state of agility comprehension coming out of school is frightening¹) that its word today is emptied of its real meaning.

But given the total misunderstanding, the denial, that reigns around it, starting to announce the return, or the strength of architecture with agility, how shall I put it? It’s true, but it’s so counter-current that I fear it might be counterproductive for them. However, that now the majority has embraced these emergent, dynamic approaches pleases me quite a bit, and I congratulate this majority of architects.

Yet the real return of architects is indeed here. It’s found in AI².

AI calls us to extend the agile technical mindset (and product because everything is connected): modularization, emergence, automation, maximum reduction of dependencies, etc. KISS (keep it simple stupid), YAGNI (you ain’t gonna need it), OAOO (one and only one) have never been more important.

This is not a return to their beliefs of old: big up front end, ultra-Cartesian design, the ivory tower. It’s much better, it’s the return of the vital importance of their skills.

The skill to prepare and maintain a framework: modularization, automation, coding rules, selection in the choice of technical elements, DDD, test automation strategy, etc. In a word, everything that frames the still wild work of AI (until when?). Framing the black boxes created by AI to contain risks. Framing, once again, means: KISS, YAGNI, OAOO, modularize, precisely not big, not massive.

Everything that ultimately frames the code. Before, we produced this code, and we defined this framework at the same time, and made it evolve, probably with major guidelines already drawn, announced convictions. Today the code escapes us, the framework regains weight. It’s not about rigidifying it, but about having real expertise on the subject. This is the role of architects.

This also means taking care of the skill development of juniors, beginners, who are to this day the left behind of AI.

This return, I’m also celebrating it by partnering with Jérémie Grodziski and David Panza in an AI venture with an end-to-end approach: from idea to impact measurement (organization, product management, architecture, development, measurements, data). Whose first steps are happening now: https://fastandslow.fr.

Long live the architects (I will deny having written this).