You Can't Prompt the Room: The Last Skill AI Won't Replace - Balázs Horváth, VisualLabs

Key Concepts

• Value-Driven Development: Prioritizing the creation of business value over the speed of code production.
• The "Bottleneck" Shift: The transition from code generation (now automated by AI) to requirement elicitation and problem definition.
• VAD (Value, Architecture, Design): A framework for ensuring software aligns with business needs.
• User Story Mapping: A visual technique for organizing user stories to understand the customer journey and prioritize MVP (Minimum Viable Product) features.
• Anti-patterns: Common mistakes in software development, such as prioritizing demos over production-ready features or measuring success by feature velocity rather than adoption.

---

1. The New Bottleneck in Software Development

Balázs Horváth argues that in the era of AI, writing code is no longer the primary constraint in the software development life cycle. Instead, the "real bottleneck" is determining what to build. Because AI can generate code rapidly, the competitive advantage shifts to "people skills"—the ability to interact with stakeholders, elicit accurate requirements, and "read the room."

• Key Insight: You can prompt an AI, but you cannot "prompt the room." Success depends on human-centric analysis rather than technical execution.
• The Henry Ford Analogy: If you ask customers what they want, they may ask for "faster horses." AI, by design, provides the most common answers (the "average"). The human role is to move beyond the average to create "cars"—solutions that represent a magnitude shift in value.

2. Methodologies and Frameworks

To ensure software delivers actual value, Horváth advocates for a return to traditional analyst toolkits, specifically:

• Story Mapping: A high-level visualization of the user journey (e.g., Contacting → Triaging → Resolving → Closing). This allows teams to identify the "backbone" of a system and define an MVP by selecting the most critical user stories.
• VAD (Value, Architecture, Design): A structured thinking process:
  1. Value: Identify what constitutes value for the customer.
  2. Architecture: Understand the underlying process and system requirements.
  3. Design: Build the system to support the identified value and process.
• The Four-Question Framework: Before building, ask:
  1. Whose problem is this? (Persona)
  2. What does winning look like? (Success metrics)
  3. What would make them refuse to use it? (Friction/Security/Platform constraints)
  4. Does it change a decision? (Impact on user behavior)

3. Leveraging AI with Structured Inputs

AI excels at pattern recognition. To get high-quality outputs, developers must provide structured, context-rich inputs:

• User Story Structure: Use the standard format: As a [persona], I need [need], so that [why].
• Contextual Documentation: Store requirements, acceptance criteria, and the "Four Questions" in a markdown file within the repository. This provides the AI with the necessary context to generate code that aligns with business intent rather than just generic functionality.

4. Identifying Anti-patterns

Horváth highlights several "red flags" that indicate a team is building the wrong things:

• The "Demo" Trap: Treating the demo as the final deliverable. A system that looks good in a presentation but isn't production-ready or used by real customers is a failure.
• Velocity Obsession: Measuring success by the number of features shipped rather than the frequency of use.
• Lack of Feedback Loops: Developing without real user testers or failing to gather feedback from subject matter experts (SMEs).

5. Actionable Insights for Teams

• Shift Talent Upstream: Move your smartest people away from pure coding and toward customer-facing roles where they can influence product strategy.
• Realign KPIs: Stop tracking "features shipped per quarter." Start tracking "features used more than twice."
• Audit Current Projects: Identify which features are being built without clear business value or user adoption.
• Mandatory Mapping: Implement a mapping session (Story Map or Business Model Canvas) before writing a single line of code.

Synthesis

The role of the software professional is evolving from a "coder" to an "analyst." While AI has commoditized the act of writing code, it has increased the importance of product management, design thinking, and stakeholder engagement. By shifting focus upstream—prioritizing the "why" and "who" over the "how"—teams can ensure they are building impactful, high-value software rather than just faster, automated versions of existing, mediocre solutions.