Embedded Files
The engineer is dead. The builder is what comes next.
Groupon CEO Dušan Šenkypl explains why the role of the software engineer is breaking down - and what it takes to operate in an AI-first company.
“The bottleneck is no longer code. It is clarity.”
The role of the software engineer is not disappearing because engineers are being replaced. It is disappearing because the boundary that defined the role no longer makes sense.
For a long time, companies were built on a separation between thinking and building. Some people defined what should happen. Others executed it. Entire functions existed to translate between the two. That separation created delays, misalignment, and a constant loss of information between intent and reality.
AI removes the need for that separation.
Once the cost of execution drops close to zero, the distinction between defining and building collapses. The person who understands the problem can now build the solution directly. The bottleneck is no longer code. It is clarity.
What matters is not the ability to write software. It is the ability to decide what should exist, and to iterate on it until it works in the real world. That is why “builder” is a more accurate description than “engineer.” The role is not narrower. It is broader.
Engineers start in a strong position in this shift, but not for the reason most people think. It is not because they can code. That advantage is already eroding. The real advantage is the way they think.
Years spent debugging systems create a habit of tracing cause and effect. Working on production systems forces decisions under constraints. Architecture work requires balancing trade-offs that do not have clean answers. This builds a way of reasoning that transfers directly into building with AI.
The missing piece is business context. Understanding customers, incentives, and behavior. That is learnable. The reverse – learning how to think in systems under real constraints – is much harder. That asymmetry matters.
Most companies are reacting to AI by increasing usage. More tools, more prompts, more output. This improves individual productivity, but it does not change how the organization operates. The same problems remain. Knowledge stays fragmented. Decisions are repeated. Context is lost between people and over time.
AI, used this way, accelerates work but does not improve the system.
The real shift happens when AI becomes part of how the company remembers. When it stops being a tool people use and starts being a layer the organization builds into.
At that point, corrections persist. Patterns accumulate. Decisions become part of the system instead of disappearing into conversations or documents. The organization starts to improve as a whole, not just through individual effort.
This is where the difference between teams becomes visible.
One of our teams ran into this early. Eight engineers working across twenty-five repositories, four domains, and multiple tech stacks. AI adoption was already high, but nothing fundamental changed. Each engineer was effectively working with their own system. Knowledge stayed local. Mistakes repeated. The system reset continuously.
"The visible effect of AI is speed. The more important effect is whether the organization learns or forgets"
So they changed the structure.
Instead of eight engineers using AI, they built a shared layer above all repositories. Rules cascade from global constraints to domain patterns to repository-specific detail. Every interaction becomes persistent. When the AI makes a mistake and someone corrects it, that correction is captured. When a pattern appears, it is encoded. The system improves with every use.
Knowledge moves out of people and into the system.
This changes the nature of the team. New engineers operate with accumulated context. The same mistake is not solved multiple times. Progress compounds. The question shifts from what someone did to what they added to the system.
That is a different model of work.
The visible effect of AI is speed. The more important effect is whether the organization learns or forgets. Most companies still forget. Even while adopting AI tools, they recreate the same knowledge every day.
The ones that win will build systems that remember.
The shift from engineer to builder is a consequence of that change, not the cause. What actually matters is whether companies continue to operate as collections of individuals, or whether they redesign themselves as systems that improve over time.
That gap will not stay small. It will compound.
We’re hiring people who build, not just talk about building. See what we’re working on and explore open roles.
Page updated
Report abuse