Everyone got excited they can suddenly code, and completely missed the point.
Coding agents showed up, software got cheap to build overnight, and a whole industry lit up: now the PMs can ship too, now the business can prototype, now everyone's an engineer. And almost nobody stopped to ask the only question that actually matters:
If we can suddenly build anything, what should we build?
Because here's the thing nobody wants to say out loud: deciding what to build has always been a bottleneck. A massive one. In most organizations it's every bit as expensive as writing the software was, it's just been hidden in the haze of software being slow and costly to produce. When delivery took months, bad product thinking was invisible. It got absorbed into the timeline. Now delivery is fast, the haze is lifting, and what's underneath is ugly.
You can pour as much AI as you want into a dev team. If the bottleneck is context trapped in a Jira ticket, it won't help. Your org is the bottleneck. And it doesn't have to be this way.
The two-line ticket
You know the dysfunction. A developer opens Jira and finds a task: one or two vague sentences, handed down from "the business." No description of the problem, no sense of who has it or why it matters. Just a thing to build, and a quiet expectation to build it.
That ticket is the symptom of an entire broken operating model. "The business" thinks, the PM translates, the developers deliver. An assembly line. And at every handoff, context evaporates.
The developer never gets near the actual human with the actual problem. So they can't think. They can't be creative about the solution, because the only information they're allowed is whatever survived compression into two sentences. And that's a tragedy, because a developer who understands the user can see opportunities literally no one else can see, opportunities that live in what's technically possible. Same as a designer sees things from their lens, and a domain expert from theirs. Cut everyone off from the user and you don't just slow things down. You amputate the entire team's ability to discover the good ideas in the first place.
This was always wasteful. Agents just made the waste impossible to ignore.
So should PMs build software with agents?
This is where the hype crowd wants a yes. And there's a version of yes I'll give you.
Coding agents are a phenomenal discovery tool. A PM who spins up a high-fidelity prototype to put something real in front of users, to learn what people actually want and will actually pay for, is using agents exactly right. That's the job getting easier, and I'm all for it.
But the moment a PM starts using agents for delivery, shipping production features, becoming a one-person delivery shop, they've gotten it backwards. Not because they'll write bad code. Because they're spending the scarcest resource in the entire company on the wrong thing.
The PM's job is the bottleneck now. "What should we build?" is the constraint everything else waits on. A PM burning their hours on delivery, when discovery is what's strangling the org, isn't being heroic. It's a prioritization failure. Discovery, not delivery. That's the whole line.
This is fixable, and most leaders don't even know it
There's a better way to work. Smart people have already written the playbook, so go read Marty Cagan if you want it spelled out. You don't have to adopt it perfectly. You don't walk into a company and swap the entire operating model on a Tuesday. It's a direction, not a dogma.
But here's what kills me: most senior leaders at companies that build software have no idea this conversation even exists. They think software development is pure delivery. Tickets in, features out, measure the throughput. They have never been told there's a way of working that gets dramatically more innovation, more value, more revenue out of the same people. So they keep optimizing the delivery line and wonder why nothing they ship matters.
If this is your every day
Then this is for you.
Stop wasting your career inside organizations that don't know any of this. It is a genuine waste of your one professional life to spend it building things nobody wants and nobody buys, in a system that won't let you get near the problem. Seek out the companies that do this well. Chase impact, not the salary ceiling. And if your job consistently has you shipping into the void, leave.
Leaders, you're meant to overhear this one. Your best people understand exactly what I'm describing, and they will walk. Fix this, or watch them go.
And to everyone who feels this every single day: there are a lot of us. Enough to make this a movement. We have the talent in Europe to build software that genuinely helps people, that innovates, that generates real revenue. We're just wasting it on broken process.
Let's stop. Let's make EU great again.
Rant over.