The Great Decoupling: Why Judgment is Eclipsing Craft in the AI Era
As we enter 2026, the technology sector is undergoing a quiet but profound transformation. The traditional “craft” of building software—writing precise lines of code, designing pixel-perfect interfaces, and drafting exhaustive product requirement documents—is being democratized by generative AI. This shift is creating a new hierarchy of value where execution is becoming a commodity, and human judgment is becoming the ultimate differentiator.
Ravi Mehta, a prominent product executive whose leadership at Tinder, Facebook, and TripAdvisor helped define the mobile era, argues that we have reached a critical inflection point. In his recent work as a product advisor and co-creator of the Reforge AI Strategy Program, Mehta suggests that the core challenge for technology professionals is no longer “how to build,” but “what to build and why.” This transition marks the end of the assembly-line era of product management and the beginning of a more intuitive, strategy-heavy discipline.
The Automation of Craft and the Rise of “Average Intelligence”
For decades, the value of a technologist was tied to their technical craft. A product manager (PM) was valued for their ability to synthesize complex user needs into a 50-page specification; a designer for their ability to master intricate layout tools; an engineer for their fluency in obscure syntax.
AI has fundamentally disrupted this equation. Large Language Models (LLMs) like GPT-4 and Claude 3.5 are now capable of producing “pretty good” versions of these artifacts in seconds. However, Mehta warns that while AI has raised the floor of quality, it has also lowered the ceiling toward a specialized kind of mediocrity. AI-generated prototypes and strategies tend toward the statistical average of their training data. They are coherent but rarely evocative or truly innovative.
The Shift in PM Competencies
Traditional Skill (Craft) Future Skill (Judgment)
Writing detailed PRDs and user stories Defining product sense and strategic “why”
Manual data collation and status reporting Interpreting patterns and emotional resonance
Fixed-role execution (The Assembly Line) Cross-functional orchestration (The Jazz Band)
Feature-first roadmapping Outcome-based differentiation and taste
Export to Sheets
This “Process Value Collapse” means that the 50/50 split between “building well” and “deciding what to build” has shifted toward a 90/10 ratio. In a world where building is easy, the risk of building the wrong thing faster than ever before has become the primary threat to business survival.
Assessing Vulnerability: The AI Disruption Framework
Not all companies are equally exposed to this shift. In his advisory work, Mehta highlights that the speed at which a company must adapt depends on its “automation risk.” Organizations like Stack Overflow or Chegg, whose core value was based on knowledge retrieval and summarization, found their use cases rapidly subsumed into horizontal AI models.
To navigate this, Mehta encourages teams to map their vulnerability across three distinct dimensions:
1. Automation Risk
How easily can an LLM replicate your core experience? If your product primarily answers questions, summarizes text, or generates basic code, you are in the “red ocean.” Differentiation must move from the output itself to the system that produces it.
2. Strategic Differentiation
A product built on proprietary data, deep human relationships, or complex network effects has a stronger foundation than one built on features alone. Trust, community, and creativity are “human speed” factors that AI cannot easily follow or replicate.
3. Customer Dependency
Products embedded in daily identity or habit are harder to dislodge. If a tool is peripheral to a user’s workflow, it can be replaced by a more convenient AI interface overnight.
From Assembly Lines to Jazz Bands
One of the most visible changes in the 2026 workforce is the blurring of traditional roles. AI is turning product teams into “jazz bands,” where the boundaries between PMs, designers, and engineers are increasingly fluid.
With the advent of “vibe-coding” and agentic tools, a PM can now generate a functional prototype before lunch, and an engineer can use AI to conduct initial user research. This overlap isn’t chaotic; it is an evolution toward “AI-native” software engineering. As execution becomes faster, the bottleneck has shifted from “engineering bandwidth” to “strategic clarity.”
“AI can generate strategy documents, but it can’t feel the market shift under your feet,” Mehta notes. “It can’t see the pattern that isn’t in the training data yet.”
This requires a new level of AI fluency from leaders. Mehta outlines a three-step process for raising this fluency within organizations:
Access: Remove friction by giving every employee access to high-tier LLMs immediately.
Expectation: Make AI use a standard part of the workflow. Instead of asking “did you use AI?” leaders should ask “how could we do this 10x faster with AI?”
Challenge: Push teams to find leverage points in prototyping and research synthesis, while protecting the human-led discovery process.
Taste as the Final Frontier
As craft becomes a commodity, “taste” is emerging as the ultimate competitive advantage. Taste is the ability to recognize what is missing, what feels right, and what users will genuinely love. It is the human intuition that prevents a product from becoming just another “average” AI output.
In 2026, the best product leaders are those who can “see around corners.” They understand that while AI can move at light speed through data analysis, the process of finding conviction and building trust still moves at the pace of human judgment. The future of technology belongs not to the fastest builders, but to the smartest orchestrators—those who know when to use the machine’s speed and when to rely on the human’s soul.
Source: https://www.atlassian.com/blog/artificial-intelligence/shift-from-craft-to-judgement-ai
Would you like me to help you apply Ravi Mehta’s AI Disruption Framework to your current product roadmap to identify your highest-risk features?
