Now it’s the same thing… just with “AI” copy-pasted on every slide.
Principal Architects running the show, somehow always in the room, rarely in the outcome.
They don’t build.
They don’t run.
But they definitely “align.”
Every problem gets a framework.
Every framework gets a deck.
Every deck gets approved.
Nothing actually gets better.
“Agentic systems,” “AI platforms,” “north star”, all discussed at length, all safely far from production.
Decisions sound big.
Ownership stays small.
Meanwhile, teams that ship are rewriting the plan in real time just to get something out the door.
At this point, the architecture isn’t enabling delivery, it’s something delivery has to survive.
But sure… one more review, one more workshop, one more “AI strategy” session.
That’ll fix it.
5 replies (most recent on top)
In our org, becoming an AI leader requires zero AI, zero engineering depth, and maximum confidence. Impressive standards.
We’ve successfully reinvented engineering:
Chief Architects who don’t architect
Principal Engineers who manage people
Leadership layers optimized for meetings, not outcomes
Honestly, the only thing scaling here is confusion.
Everyone is now an ‘AI leader’ - which is fascinating.
EA is a mirage other than clearpath compliance ( mostly driven by risk) not seen any value from them.
Expect them to recreate discover(CHI) team here and struggling to understand the businesses which are not cards/payment.
Frameworks are supposed to be reused for similar problems. They produce ppt, it’s not even close to a framework .
A pattern worth questioning: architecture leadership cycles where momentum stalls for a year, then gets reset with familiar hires from prior orgs.
Instead of building continuity and accelerating delivery, we end up rebuilding influence networks and rediscovering the same structures under new branding.
Across the board, it raises a simple question, are we optimizing for system outcomes, or for organizational familiarity?
Take a look at Paul’s org under architecture leadership, if this is the blueprint, it’s a masterclass in turning simple problems into complex ones while delivery slows to a crawl.