Bal!zs Horv!th

The Shift from Syntax to Stakeholders Let's analyze where the real work of software engineering is moving. Bal!zs Horv!th, founder of VisualLabs, points out a stark reality: writing syntax is no longer the bottleneck. When large language models generate clean implementations on demand, the most expensive part of development moves upstream to requirement gathering. You can write perfect prompts for an artificial intelligence, but you cannot prompt a room full of conflicting human stakeholders. Success depends entirely on human-to-human interaction. Why AI Produces the Fastest Average During a recent internal hackathon, VisualLabs built 21 functional agents. They quickly killed 17 of them. The technology worked, but the tools solved no real problems or lacked essential data access. AI models deliver the most common answer by design. Without precise, human-led engineering of requirements, you get a digital equivalent of a faster horse instead of a car. Teams must use classic story mapping to structure the persona, the need, and the underlying logic. This structure works because systems excel at recognizing these traditional patterns. The VAD Path Over the Demo Trap To build tools that actually run in production, teams must follow the Value-Architecture-Design (VAD) sequence. Always start with value: whose problem are we solving, and what does winning look like? Next, map the architecture that supports the data flow. Only then do you design and write the code. Skipping straight to code creates the demo trap—shiny applications that look impressive in meetings but get abandoned immediately because they fail to change real-world decisions. Auditing the Wrong Metrics We must change how we track engineering success. Measuring the sheer volume of features shipped is a dead metric. Instead, look at frequency of use. Track how many features customers actually run more than twice. Move your smartest technical minds away from pure coding and place them directly in front of clients. Building has become cheap; knowing what to build is the new competitive moat.