A waterfall dream
By Christos Gkoros
- 3 minutes read - 628 words
The shape is contracting
Product engineering got complicated for good reasons. Product managers to translate intent. Designers to translate experience. Engineers to translate constraints. Each role existed to own a part of the feedback loop process because no one person could hold all at once. At least not at scale.
That is the assumption AI is dissolving. The translation layers between roles were never the work — they were the cost of distributing it across people.
The shape is contracting back to something simpler. Not because we got better at coordination. Because we need less of it.
Waterfall, but rapid
The agile era was a response to a real problem. We could not afford to spend too much time thinking about what we wanted to build, only to spend months learning that we had built the wrong thing. Requirements drifted, handoffs were slow, and the cost of being wrong was crushing. Short cycles, constant feedback, and parallel exploration paid for themselves.
That trade-off changes when one person, with AI, can spec, design, build, and ship in a week. The handoff cost collapses. The drift cost collapses with it. Long-running parallel tracks stop being necessary — you can just do it linearly and finish before you would have finished the standup.
That is practically a waterfall. Plan it. Design it. Build it. Test it as a whole. Release it. This pattern returns because all the phases compress, not because we re-learned to love big design up front.
The builder archetype
The teams I have worked on have always valued people who could understand the bigger picture. Engineers with a business mind. Designers who knew JSON and REST. PMs who could prototype or spot performance constraints. T-shaped people. The flat side of the T was the cost of admission, and most of the org chart existed to compensate when it was missing.
AI fills the flat side. The engineer who could not do user research now has a research synthesizer. The designer who could not create web pages now has a code generator. The PM who could not evaluate technical assumptions now has an engineering consultant on their side.
The builder archetype is one person carrying a product end-to-end because the gaps that used to require other humans are now filled by tools. Not because the rare profile got more common. Because the gap got smaller.
Fill versus assist
There are two things AI is doing here, and they are not the same.
It fills gaps where the person lacks the skill — the engineer who does not write copy, the PM who does not write SQL. The output is good enough, and good enough is enough, because the alternative was nothing.
It assists where the person already has the skill — the engineer designing the architecture, the designer working through the flow. The output has to be above a higher bar, because the person can tell when it is wrong.
Most of the productivity gain comes from the first. Most of the quality gain comes from the second.
The new shape
The new shape is not a squad full of translators. It is a small stream of builders owning planned features end-to-end, with AI-assisted QA around them. Linear cycles, because the cycles fit in a week. AI fills in the gaps in specialization, and humans take care of the parts that need taste or accountability.
The org chart becomes simpler. Fewer hand-offs to coordinate means fewer roles whose job was the hand-off. The headcount that survives is the headcount that owns outcomes, plus the operations weight that does not go away once anything ships at scale.
It will not feel like progress to everyone, but it will get companies where they need to go in a brutally straightforward way.