For more than two decades, the software industry has evolved through a series of transformative waves—from desktop applications and web platforms to cloud computing, mobile-first architectures, and Software-as-a-Service (SaaS). Today, however, we are witnessing a shift that may surpass all previous technological transitions: the emergence of AI-assisted software development as a new paradigm for building digital products.
The impact is already visible. Development teams are becoming more productive, repetitive engineering tasks are increasingly automated, and software can be generated at a speed that would have seemed impossible only a few years ago. At the same time, fundamental questions are emerging. What happens to junior developers when AI can write most of the boilerplate code? How do engineers build expertise if they rely on generated solutions instead of understanding the underlying principles? Will software engineering become a profession of prompt creators, or will deep technical knowledge remain the ultimate differentiator?
To explore these questions, I spoke with Ilija Mishov, one of the most experienced technology leaders from our region. Throughout his career, Ilija has been at the forefront of software engineering, technology leadership, and digital transformation. He has served as Founding Partner, Technology, and CTO of IT Labs and today serves as Founder and Strategic Digital Transformation Advisor at Tektopia, where he helps organizations modernize their technology landscapes and navigate complex transformation initiatives.
Over the years, Ilija has worked with companies ranging from startups to large enterprises, leading software delivery, organizational growth, and modernization efforts. His expertise is particularly relevant in a period when many organizations are simultaneously facing two challenges: adapting to the AI era while bearing the burden of legacy systems designed for a completely different technological reality.
In this conversation, we discuss the future of software development, the role of AI in engineering teams, the evolving expectations for software professionals, and why modernization has become a strategic necessity rather than a technical option. We also explore the future of software services companies, the changing economics of software development, and the opportunities that remain for those willing to embrace the next generation of technology.
If you are a software developer, technology leader, entrepreneur, or business executive trying to understand how AI will reshape your organization, Ilija’s insights offer a valuable perspective on what lies ahead—and on the skills, mindset, and strategic decisions that will matter most in the years to come.
VBU: What’s your opinion about the new software development paradigm that deeply integrates AI?
IMI: AI has found its purpose in software development, and we cannot ignore the performance boost, especially in repetitive tasks like CRUD operations, writing unit tests, etc. But beneath this efficiency, I’m seeing a much more significant shift: the developers are becoming prompt engineers. I see a new generation of engineers spending more time constructing the perfect prompt with the fewest tokens than diving into the code, the execution logic, or the architecture. This is both a positive and a negative. I’m not a fan of having fewer developers who think about elements like clean code, code optimisation, correct iteration logic, etc. Even though we now have tools that can help with these tasks, it often seems like, as developers, we’re losing one of the most important responsibilities – taking care of what we are producing. Now, many aren’t really producing anything; they’re just correctly asking for it to be produced.
And, regarding the development paradigm, there is also another aspect. So far, AI can help in building the solution, but it cannot deliver it. This still leaves some room for roles like QA engineers. I know that a good chunk of QA is also automated and AI-ed as well, but, in the end, the software solutions are mostly used by humans, thus they must be tested by humans. Somebody has to do at least a sanity check.
There is also a third segment to all of this, and that’s the social and professional interaction we’re slowly losing. I remember that 25 years ago, it was very difficult to find any knowledge on the internet to solve a problem, so we gathered in hallways, kitchens, or other company spaces to discuss problems and find solutions. The only “knowledge-sharing” we had access to was conferences and other major gatherings, but none of those could really help you with your daily tasks. Then, the internet expanded, so the gatherings happened on Stack Overflow, Reddit, X (ex-Twitter), and other similar networks. All of this is almost gone now. If anyone has a problem, they ask their preferred AI engine. There is no discussion, no brainstorming… nothing. If the person is not critical enough, they will just accept whatever the AI engine outputs without any question. This is not the way one should learn and evolve.
VBU: What do you believe the young software developers should learn, or is this seniority already obsolete?
IMI: It may seem strange, but I think it’s very simple: if we don’t have juniors because AI took over their job, we will soon be lacking seniors as well. And it becomes progressively harder for young developers because it’s more challenging to even get a working chance and start building their experience. They can focus on building quality AI prompts, but this way, they will never learn the basics, and their further development will be much harder and full of knowledge holes. I am an old school guy, and I truly believe in basics and progressive learning. Even before the AI boom, developers would struggle with a task and just copy/paste something they found online to solve the problem. And they did, but when you ask them to explain the solution they’ve built, it became very clear that they don’t understand it at all. How could I expect them to build something on top of that code if they don’t understand it completely? AI now amplifies this. So, my advice to young developers is: use and learn AI because you must stay up to date with the technology, but explicitly examine and understand every piece of code and each line the AI engine produces. Think about what it does, why it is there, whether it can be optimized, reused, etc. Don’t take its outputs as gospel.
VBU: In an interview with Anthropic co-founder Jack Clark published on June 5, 2026, he said that the popular coding chatbot Claude runs on code that the system wrote itself in 80%. His approximation is that the entire code (100%) could be written by the system in 2 years. Is it worth dreaming about becoming a software engineer/developer nowadays if you are young and facing the challenge of deciding where to focus your time and energy to build your future career/professional life?
IMI: The short answer is YES. The longer answer is a bit more complicated. In the past 20 years, as the demand for software engineers grew rapidly, the quality of those software engineers dropped significantly. The companies tried to present their workforce as more competitive on the market, and they started assigning titles to their employees with no real background or credibility. For example, one can now become a senior engineer with only 5 years of experience and with working only on a single project for their entire career until then. Maybe they have achieved seniority on that particular project, but still, their maturity level is way too low for their title. Now, with the help of AI, this will change. The engineers will still be needed, but not as much as previously. Somebody will have to build new products. Somebody will have to build new AI engines, robots, services, etc., but the quality bar for engineering capacity will be raised much higher than it was before.
VBU: I’ve been spotting your posts on LinkedIn lately. You posted about legacy software systems and ways to tackle potential issues an organization may face with old systems. So, what are your thoughts about it?
IMI: Legacy systems are a universal reality, and especially now that AI is taking over. If we expected that the system we bought (or built) could be utilized for the next 10 to 15 years, that utilization period is cut down to half or even more nowadays. In the nature of a good and progressive business, it should always evolve and adapt to the new era. However, companies, especially those outside the ICT industry, are struggling to keep pace and are blocked from different perspectives. It’s almost like they got too accustomed to using the solution and cannot imagine changing or upgrading it, or they are too scared of the downtime or the data loss that might happen, or having the new system might change the processes they have already established, etc.
These are valid fears, but refusing to do nothing is an even bigger risk that makes the organization stagnant. In general, this is noticeable through the products and/or services they provide, and how they keep up with their competitors. The larger the organization, the harder it becomes. I have been dealing with problems like these almost all of my professional career, and even now with clients, and my vision is to empower the company leadership to break through their fears, to accept the challenges of modernization, and leave me the heavy lifting of the process.
VBU: What’s your bet on bespoke software development companies and outsourcing agencies that were leaving their golden years before 5-10 years? What’s their business potential in the future, considering AI speeds up things, and marketing tries to sell hundreds of tools with the narrative “You don’t need to pay tens and hundreds of thousands of dollars to build your software and SaaS.” After decades of experience in the bespoke software development industry and as an ex-co-founder of IT-Labs, would you start a new software development company/studio/agency if you had the chance now?
IMI: There will always be non-ICT companies with a good idea about improving their service, product, customer reach, etc. These companies, instead of building their own ICT team, will reach out to ICT companies for help. However, the main product of the service-oriented ICT companies is the “man-hour”. This is how they build their revenue. Today, with all of the SaaS products out there and the help of AI agents, the quantity of “man-hours” an ICT company can sell is significantly decreased. This leads to building a much larger client portfolio to retain the revenue as it was, so their focus should be on increasing quality and performance in order to endure. Now, regarding whether I would start a new software development company today, I would, but not a service-oriented one. Now, I am more focused on products. Not pure software SaaS products, but multidisciplinary products that will help other non-ICT industries get on top of the new AI era.
Be the first to comment