Strategic Taste and Problem Identification Over AI Execution

The idea that AI will turn every worker into a software developer is a persistent, seductive myth. While AI lowers the technical barrier to writing code, it does nothing to bridge the fundamental gap in identifying problems, navigating organizations, and applying the rare, non-obvious taste required to design effective systems. The real bottleneck to innovation is not the ability to build; it is the ability to see a problem, articulate a solution that breaks from the status quo, and manage the human and organizational friction required for adoption. For leaders and practitioners, the competitive advantage lies not in adopting every new agentic tool, but in cultivating the specific, difficult-to-replicate skill of identifying high-leverage problems that others are too busy to notice.

The illusion of the grassroots architect

The recurring narrative in technology, from Microsoft Access to the no-code movement and now generative AI, is that we are on the verge of a grassroots revolution where every employee builds their own bespoke solutions. Benedict Evans argues that this ignores the reality of how work actually gets done. Most professionals are optimized for their specific domain, such as sales, law, or graphic design, not for system design. They are busy executing their craft, not auditing their workflows for automation potential.

Even when a problem is visible, the leap from "this is tedious" to "this is a software-solvable process" is vast. As Evans notes, the mark of great software is the realization that a problem can be reframed into a new, more efficient process. This requires a specific type of taste, a deep, intuitive understanding of how workflows, data, and human incentives intersect.

"There's a huge difference even between seeing the problem and being the right person to work out like what should the processes and the workflows and the networks and the functions be that would make this great."

-- Benedict Evans

The hidden friction of institutional adoption

Even if an individual manages to build a functional tool, they hit a wall that has nothing to do with code: the 50-person company problem. Enterprise software is rarely a solitary endeavor. It exists within a web of compliance, security, and cross-departmental dependencies.

The transition from a 10-megabyte Excel file, the ultimate shadow-IT tool, to an institutionalized system like SAP is not just a technical upgrade. It is an organizational one. It requires someone to take responsibility for the system, ensure its security, and manage the inevitable friction of changing how a team operates. Most bottom-up tools fail because they never solve the problem of getting the other 49 people to adopt them.

"There’s a point where you do an excel and there’s a point where you buy a thing because you want to know that somebody thought about how it should work and somebody is responsible for it and will fix it, and there will be bug fixes and security."

-- Benedict Evans

Why taste is the new moat

In an era where AI can generate boilerplate newsletters, code, or legal drafts, the value of standard output is collapsing. If everyone uses the same tools to produce the same average-quality work, the output loses its competitive edge.

The implication for the professional is clear: your value is no longer in the execution of the task itself, but in the unique opinion you bring to it. Whether you are a lawyer or a strategist, the ability to have a stark opinion that deviates from the conventional path is what creates value. AI can provide the facts, but it cannot provide the strategic taste required to build a tool or a process that actually moves the needle.

Key action items

• Audit for shadow IT (Immediate): Identify the mission-critical Excel files or manual workarounds currently running your team. If they are essential, treat them as high-priority candidates for formalization, not just AI-assisted automation.
• Shift focus from building to problem-finding (Ongoing): Stop asking "How can I use AI to build this?" and start asking "What is the specific, tedious process in our workflow that no one else has bothered to re-engineer?"
• Develop your product taste (12-18 Months): If you are a domain expert, spend time studying how software products are built in your field. Understanding the modalities of software will allow you to articulate requirements to AI agents that others cannot.
• Prioritize adoption over invention (Quarterly): When proposing a new tool, spend 80% of your effort on the social and organizational aspects of adoption, such as permissioning, compliance, and stakeholder buy-in, rather than the technical build.
• Cultivate a unique perspective (Ongoing): As AI commoditizes the how of your work, double down on the why. Your competitive advantage is no longer your ability to generate standard results, but your ability to apply a unique, informed opinion to complex problems.