In 2026, the definition of a “good developer” is undergoing one of the greatest tests in the industry’s history. Given that artificial intelligence can generate correct code in a matter of seconds, an engineer’s value no longer stems merely from their knowledge of frameworks.
Today, something else is becoming critical: the ability to design solutions, break complex problems down into the right components, and critically evaluate what the machine suggests.
Is AI truly taking developers’ jobs? How is the role of leadership changing? And what does this mean for juniors?
Without any doomsday fearmongering or hyper-optimism, I discuss these topics with @Marcin Zięba, Engineering Director at Ocado Technology.
Table of Contents
Is AI the end of the developer profession?
Bartek: Media articles crop up constantly predicting the end of yet more professions due to be replaced by artificial intelligence. Do these forecasts carry any weight? Is AI truly a threat to software developers?
Marcin: I think the rumors of the developer profession’s demise are rather exaggerated. I recently read an insightful interview with Sundar Pichai (CEO of Alphabet), who views artificial intelligence as an accelerator for productivity and innovation. I agree with him on this point. AI gives us the opportunity to greenlight projects that, until recently, were shelved for being unprofitable or too time-consuming. Companies will want to launch far more of them, which, paradoxically, will increase the demand for talent.
However, we must honestly admit that the role of the developer is shifting heavily under the influence of AI. What’s more, this transformation is happening incredibly fast.
How is the engineer’s role changing in an AI-Driven world?
Bartek: How do you assess the maturity of current AI tools? Can they be trusted with production tasks yet?
Marcin: While a year ago we looked at an AI model as a talented junior who needed their hand held, today it is much closer to a senior-level partner. Thanks to large context windows, models can “see” the entire project, spot architectural patterns, and generate cohesive solutions.
However, I would like to draw a sharp line here: it is a very specific type of senior. It possesses immense technical knowledge, but zero business intuition and no sense of consequence. It is still a powerful machine, not a human being.
Because of this, the developer’s role is shifting from a “craftsman” who writes code to a “conductor” and verifier. Artificial intelligence can handle most of the grunt work, but it is the human who must define the goal, break it down into meaningful tasks, and – most importantly – take responsibility for the outcome. A model will generate a solution, but it won’t take an on-call shift for a developer, nor will it answer to the board for a production outage.
How teams respond to AI – from resistance to standard
Bartek: How are the developers in your organization reacting to these changes?
Marcin: Resistance to the new is natural, but at Ocado, our approach to AI is highly pragmatic, which stems from our DNA. For years, we have been building “smart” features – utilizing advanced algorithms and machine learning to optimize logistics, power shopping recommendations, or control the robots in our warehouses. For our engineers, working with intelligent algorithms is standard practice, not a novelty.
What has changed recently, however, is that we have given this a strategic framework. We treat AI as a priority both at the company-wide level and within my department. We created a concrete strategy that outlines two clear directions of action: supporting product innovation and improving the “developer experience.” We want AI to take tedious, repetitive tasks off developers’ plates and support them in making design decisions.
Thanks to this approach, the discussion within teams isn’t about whether it’s worth it, but how to use it wisely.
Bartek: And that is a crucial point from a recruitment perspective. We are increasingly seeing that a mere list of technologies on a CV is no longer a sufficient indicator of quality. Today, it is far more important whether an engineer can take ownership of a product and understand the business context of their decisions. You can’t verify that with a simple framework proficiency test.
Marcin: Exactly. This is happening because the barrier to entry into a specific technology has lowered significantly – especially for people with extensive general experience.
The best proof of this is the fact that in our company, managers are returning to working with code. This is a group that has a great understanding of the business but had moved away from daily coding over the years. AI allows these individuals to get back in the game. It lifts the burden of fighting with syntax or knowing libraries inside out, allowing them to dive deeper into the project. As a result, leaders are once again closer to the problems being solved.
In this context, skepticism toward AI in 2026 is no longer “healthy caution” but a lack of a core competency. It’s a bit like insisting on digging with a shovel when there’s an excavator sitting right next to you – you can still work that way, but you are drastically limiting your capabilities.
That is why, in order not to leave people facing this challenge alone, we have focused on systemic education. We encourage training – including in “spec-driven development” – and provide access to professional tools like Cursor or Copilot, while continuously developing the knowledge base for our internal AI assistant.
The role of the Technical Leader in the AI era
Bartek: Since the barrier to entry into technology has lowered, where is the weight of a leader’s responsibility shifting today?
Marcin Zięba: I mentioned leaders diving deeper into projects. Today, this has already become a standard expectation in most organizations. We see value in a leader genuinely understanding the matter their team is working on. Thanks to this architectural knowledge, the leader becomes a true partner for the business – they are able to answer stakeholders’ questions on the fly, instead of ending every conversation with the phrase: “I’ll get back to you, I just need to check with the team.”
The shift we are experiencing is happening at record speed, and for many people, it is simply overwhelming. Above all, a leader today must provide support during this process: they must look out for the team’s well-being and guide the wise development of competencies, ensuring that no one feels left behind by the breakneck pace of technology.
Bartek: Looking out for well-being and development sounds like a purely “human” challenge, but I imagine technology can practically support leaders here. How does this translate into concrete actions within the team?
Marcin Zięba: At the single-team level, it can mean encouraging people to use AI for daily tasks, implementing automated code reviewers, or building knowledge bases from your own documentation to feed a local model or a company RAG system. These are tangible benefits that save time, for instance, in support workflows.
On the other hand, at the multi-team level, the manager’s weight of responsibility shifts from “delivery management” to ecosystem design. It becomes crucial to build a culture of innovation where there is permission to experiment – including failures – and space to exchange knowledge. I operate on the assumption that if a team doesn’t “fail forward” when trying to apply AI, it means they aren’t learning fast enough. But to do that, a safe environment is absolutely essential.
Bartek: What about security? Many companies are afraid of so-called shadow IT, meaning employees using tools on their own initiative.
Marcin Zięba: This is a serious challenge. Naturally, the foundation must be commercial licenses that protect our intellectual property and ensure GDPR compliance. We must continuously educate our people and sensitize them so they understand why they cannot paste company code or data into public, free tools.
Conversely, shadow IT is often a signal that people have found a better, faster way to work—the company just hasn’t caught up yet. Instead of fighting grassroots initiatives, a leader should “civilize” and scale them. As technical professionals, we have the competencies to assess the risk and formulate a recommendation for the company.
A prime example from my department is the use of a local Whisper model. It supports our UX researchers in analyzing recordings from user sessions. Automated transcription saves them an immense amount of time that used to be wasted on manual note-taking. Because the model runs locally on the laptop, the data never leaves the company and remains secure. In this way, we turned a potential “grey area” into an official, secure working standard.
What competencies are critical for a Developer in an AI-driven world?
Bartek: What competencies should IT specialists be focusing on developing today?
Marcin Zięba: We are moving into a mode of delegating tasks to machines, which is why precision in formulating thoughts is becoming critical. If an engineer cannot break a problem down into logical steps and provide the necessary context, the AI will generate working code… that solves the wrong problem.
The second aspect is technical intuition. We must remember that models are not thinking entities that “understand” architecture in a human way, but rather advanced statistics. We have to rely on our own technical intuition. It allows us to assess whether a model is hallucinating or proposing an unsafe solution. This intuition is what protects us from so-called “vibe coding” – the uncritical acceptance of code simply because, at first glance, it “looks good.”
Bartek: “Vibe coding” usually has a positive connotation – anyone can build their own app and validate an idea in practice, even without deep technical knowledge. Do you see a danger in this?
Marcin: Indeed, you can create great prototypes or Proof of Concepts this way. That has immense value – the barrier to entry drops, and anyone can quickly check if their idea will catch on. However, the problem arises when we treat such code as production-ready. Solutions generated on the fly by AI rarely meet standards for security and scalability.
A prime example is the story covered in the media just a few days ago about Mr. Dawid and his KSeF (National e-Invoicing System) integration app. Built with the help of AI, it worked perfectly, and the delighted author shared it with the world – inadvertently exposing critical company data in the process. Perhaps haste and the desire to show off the solution before anyone else played a part, or perhaps it was simply a lack of experience.
How AI helps pay down technical debt
Bartek: At the beginning of our conversation, you mentioned the interview with the CEO of Alphabet, who spoke about artificial intelligence as an accelerator opening up new possibilities. Do you have concrete examples of tasks that went from being “shelved” to actually completed thanks to AI?
Marcin Zięba: I have a great recent example. We are currently wrapping up rewriting one of our applications to a different tech stack. It was a classic piece of technical debt. The application works fine, but it handles a secondary functionality, so the project kept getting pushed back – there was always something more urgent.
One of our managers took on the task. Thanks to AI, he rapidly got up to speed on a new UI framework that he hadn’t known before. He didn’t write the code himself, but used his experience to ask the model the right questions about architecture and potential migration pitfalls.
As a result, we got a realistic feasibility assessment and an action plan. This allowed us to prioritize the work and hand it over to a specialist who, with our support, delivered the project exactly as planned.
The fate of juniors in an AI-driven world
Bartek: In a world where AI has reached a senior level, is there still room for juniors?
Marcin Zięba: That is a very good question. You are touching upon a risk that has been discussed loudly for months. I believe that as an industry, we must actively eliminate this risk. I absolutely do not think that AI will replace juniors. If we stop hiring entry-level developers, in a few years we will wake up with a massive skills gap and a lack of specialists who understand our systems. Today’s seniors will grow, get promoted, change companies, or retire. This talent renewal process is vital for the survival of businesses.
Fears that AI will take jobs away from juniors stem from the fact that during the pandemic, IT was hailed as an “Eldorado,” which attracted crowds of people hoping for a quick career pivot via “bootcamps.” Individuals who entered IT but didn’t deepen their knowledge are indeed in a difficult position today, because AI is taking over the baseline writing of code.
Today, the bar is being raised. We are returning to the roots, where solid engineering fundamentals matter—algorithms, data structures, understanding how systems work. This is the baseline that separates someone who merely writes code from an engineer who solves problems. For these juniors—those who are curious and understand why something works under the hood—there will always be a place.
Bartek: It’s essentially a market reality check—we are going back to basics. How, then, do you view the potential of today’s graduates, who have treated AI as a standard tool since their first year of university?
Marcin Zięba: Yes. This reality check filters out accidental people, but on the other hand, we now have access to juniors I would call “AI-native.” These are individuals who already use artificial intelligence during their studies, learn machine learning, and train their own models as part of their master’s theses. This technology is a natural tool for them, not a threat. We are going to need these people.
The challenge remains in developing the technical intuition I mentioned earlier. We cannot allow juniors to become mere “prompt operators.” When it comes down to it, teaching methods here remain old-school. It is still a master-apprentice relationship and close collaboration with seniors. What does change is the toolkit: a senior no longer has to teach a junior syntax, but they must teach them to read between the lines of AI-generated code and spot errors where something looks seemingly correct.
As managers, we must create space for this. Succumbing to the narrative that seniors can just “handle more” on their own with AI to optimize costs is short-sighted.
Bartek: What would be your advice to software developers in 2026?
Marcin Zięba: Do not treat AI as a threat, nor as a magical solution to all problems. It is simply another powerful tool in your kit. Experiment without fear, but maintain your critical thinking. Let AI speed up your work, but never stop deeply understanding what you are doing.
The combination of artificial intelligence and your own technical intuition will be your greatest asset on the market.
How to build IT teams in the AI era: Key takeaways for companies in 2026
A clear conclusion emerges from my conversation with Marcin: paradoxically, in the AI era, it is the “human” engineering competencies – such as breaking down complex problems into their core components, critical thinking, and product ownership – that are becoming the hardest currency on the market. AI does not lower the bar; it shifts it toward architectural planning, security, and the scalability of the entire system.
At itMatch, we know that finding an expert in a specific framework is only half the battle today. The real challenge is reaching engineers with “technical intuition” – those who can wisely oversee the work of AI and verify that the generated solutions are truly production-ready.
If you want to discuss how to effectively recruit these professionals today and build teams ready for this new space of opportunities, please feel free to reach out.
About the authors:
Interviewer: @Bartek Toporkiewicz – CEO of itMatch, an expert in building strategic recruitment partnerships and scaling IT teams based on hard market data.
Interview Guest: @Marcin Zięba – Engineering Director at Ocado Technology, a leader implementing modern, AI-driven engineering models.
