Balázs Horváth emphasizes that while AI simplifies coding, the irreplaceable human skill lies in understanding and defining what to build through strong stakeholder engagement and effective use of analysis tools like story mapping. He advocates for a value-driven development approach focused on real user needs and business goals, warning against prioritizing feature output over meaningful adoption and urging teams to leverage human insight to guide AI-assisted software creation.
Balázs Horváth discusses the evolving role of human skills in software development amid the rise of AI. He emphasizes that while AI has made coding easier and less of a bottleneck, the critical challenge now lies in determining what to build. This requires strong people skills and the ability to engage stakeholders effectively—something AI cannot replace. Horváth illustrates this with an internal hackathon example where most AI-generated ideas failed to deliver business value, highlighting the importance of understanding real needs before development.
Throughout his 13-year career bridging business and IT, Horváth has focused on eliciting clear requirements to create valuable software. He notes that although interaction with systems and AI is changing rapidly, the fundamental process of understanding customer needs remains constant. The key skill now is mastering analysis tools such as story mapping, business model canvas, and value canvas, which help teams visualize user journeys and prioritize features that deliver real value.
Horváth delves into story mapping as a particularly effective technique. By breaking down user processes into stages and user stories, teams can identify minimum viable products (MVPs) and build incrementally. He stresses the importance of writing user stories in a structured format that AI can understand and use effectively, enabling better collaboration between humans and AI in software design. This approach ensures that development efforts align closely with user needs and business goals.
He introduces the VAD (Value, Architecture, Design) framework to guide the development process, starting with understanding whose problem is being solved, what success looks like, potential barriers, and how the solution influences decision-making. Horváth argues that this method is essentially good product management but is more critical than ever because everyone now has access to similar AI tools. The differentiator is the ability to deeply understand and articulate business needs to guide AI-driven development.
Finally, Horváth warns against common pitfalls such as focusing on feature velocity over adoption, treating demos as deliverables, and neglecting real user feedback. He advocates shifting the smartest people toward customer-facing roles to better define what should be built, as building has become relatively cheap. He encourages teams to audit their current metrics, prioritize meaningful usage data, and incorporate user story mapping before development to ensure they build the right solutions, not just more solutions.