How the 10th edition of Hack9 proves AI-first is no longer a strategy it's a habit
Ten editions is a long time in technology. Long enough to watch an industry go from nervously mentioning AI in a meeting to spending an entire Friday building with it – openly, competitively, and with genuine craft.
That is what happened on May 22nd, when more than 100 Levi9 engineers across Amsterdam, Belgrade, Iași, Kyiv, Lviv, and Novi Sad gathered for a full day of AI-first software development on the 10th edition of Hack9. Thirty-five teams. Four and a half hours. One theme: Code less. AI more.
It is worth pausing on that theme for a moment, because it was not chosen as a catchy tagline. It was chosen as a statement of direction. Not “explore AI” or “experiment with AI.”
Code less. AI more.
As in: this is the way we are building now, not the way we might build someday.
That shift, from experimentation to habit, is the most important thing Hack9 2026 demonstrated. And it has implications that go well beyond a single company event.
Why Professional Software Engineering Matters More in the Age of AI
At the opening of Hack9, Levi9 CTO Jan Dolinaj framed the issue clearly: AI is making it easier for more people to build software, but ease of creation is not the same as quality of delivery. The harder challenge, and the more valuable one, is building software that can scale, hold up, and perform in real conditions. That is the role of professional engineering.
This is a counterintuitive argument, and it is exactly right. The arrival of AI tools has made it easier than ever to generate code, create prototypes, and ship something that works on a demo. What it has not made easier is building software that is secure, maintainable, well-documented, performant under real-world conditions, and aligned with actual business requirements over time.
That requires engineering judgment, the kind that comes from experience, from standards, from a team that has been developing and refining its practices over years.
The companies that will struggle in the next few years are not the ones that failed to adopt AI. They are the ones that adopted it carelessly, vibe coding their way to a demo and discovering in production what professional engineers would have caught in planning. The companies that will thrive are the ones with partners who already know how to use AI well, not just how to use it.
What AI-First Software Development Actually Looks Like in Practice
For the past few years, AI-first has been a phrase that mostly described an aspiration. At Hack9 2026, it described an operating reality.
The event itself was structured around a set of principles that would have seemed aspirational even two years ago. Teams were expected to build artifacts, real, usable, composable tools, not just prototypes or demos. They were evaluated not on novelty but on clarity, reusability, documentation, security, and real-world value. The judging criteria were applied by AI agents, with two humans, Principal & Solution Architects, in the loop to evaluate impact and originality.
That last detail is worth sitting with. The jury that evaluated 35 teams’ AI-built solutions was itself partially composed of AI agents. Not as a gimmick, as the most consistent and scalable way to apply a rigorous, consistent rubric across that many submissions. The humans in the loop were there for the things that AI does less well: contextual judgment, originality, genuine impact assessment.
This is what mature collaboration between humans and AI looks like: AI supports scale and speed, while humans remain responsible for judgment and decision-making.
10 Years of Hack9: What a Decade of Practice Produces
The value of a ten-year Hack9 is not the anniversary itself, especially when the real magic number was nine and we have already passed it. It is the accumulation of institutional knowledge that ten years of doing something produces.
Levi9’s technology community has been running Hack9 long enough that the conversations happening inside it are not “what is AI” but “how do we use it better.” This year’s three pre-hackathon workshops covered AI across the entire software development lifecycle (SDLC), the architecture of skills and agents, and token-efficient prompting strategies that reduce cost without reducing quality. These are not introductory topics. They are the questions of practitioners who are already operating in this environment and want to operate more effectively.
The solutions built on Hack9 day reflected that maturity. Teams built security threat modelling systems, automated proposal generators, technical debt scanners, on-call support tools, framework migration pipelines. These were not proof-of-concept toys. Several were built on real projects, with real data, solving problems the teams had encountered that week.
What CTOs Need to Ask Their Software Development Partners Right Now
If you are a CTO or technology decision-maker reading this, the relevant question is not whether AI will change software development. That is already settled. The relevant question is Does your software development partner already know how to work this way?
The gap between a team experimenting with AI tools and a team that has built genuine AI-first software development practices is significant, and it compounds over time. That gap shows up in four measurable areas:
Delivery speed. Teams with established AI-first practices consistently reduce development cycles. Teams have reported up to 4x faster feature delivery compared to traditional approaches.
Code quality. Embedded security guardrails and automated review agents catch issues throughout the development cycle, not just at the end. The result is fewer vulnerabilities reaching production and faster remediation when they do.
Onboarding time. When knowledge lives in structured documentation and defined agent workflows rather than in people’s heads, new team members contribute meaningfully within days rather than weeks.
Project predictability. Defined agent architectures, cost management strategies, and consistent processes reduce surprises — in timelines, in budget, and in production.
Hack9 is, among other things, an annual proof point that Levi9’s engineers are not waiting to be told how to work in the AI era. They are already showing what leadership in this space looks like.
The right question to ask a software development partner is not “do you use AI?” Almost everyone does. The right question is: do you have defined standards for AI-first software development? Can you show me how your SDLC is structured around it? What security guardrails are built in throughout the process, not just at the end? The answers will tell you whether you are looking at a team that has adopted AI as a habit, or only as a headline.
