The same capability does not stay the same kind of work forever.
That is one of the most useful ideas in Wardley Mapping. Capabilities evolve. They begin in genesis, where the thing is new and uncertain. They become custom, where experts build and adapt. They become products, where the market understands the need and vendors compete on features. Eventually many become commodities or utilities, where users expect reliability, low cost, and standard access.
Operators get into trouble when they manage a capability according to what it used to be.
A decade ago, using large-scale machine learning in many workflows required rare talent, custom infrastructure, and deep experimentation. Today, model access is widely available. The frontier is not gone, but part of the stack has moved. Some teams still talk as if every AI decision is research. Others talk as if every AI problem has become plug-and-play. Both are wrong because they flatten evolution.
A map lets you separate the layers.
In an AI-native stack, model APIs may be productized. Cloud infrastructure is mature. Identity and access management should be boring. Prompt practices may be moving from craft to productized tooling. Evaluation may be uneven. Domain-specific context assembly may still be custom. Agent governance may be early. Workflow redesign depends heavily on local operating knowledge.
If you manage all of that as innovation, costs explode and reliability suffers. If you manage all of it as procurement, learning dies.
Evolution changes what good looks like.
Genesis work needs exploration. Small bets. Fast learning. Direct contact with users. Low ceremony. Clear kill criteria. You are trying to discover whether something is real.
Custom work needs skilled builders close to the problem. It requires judgment, feedback, and careful fit. You are shaping something that cannot yet be bought cleanly.
Product-like work needs comparison, integration, vendor assessment, feature tradeoffs, and adoption discipline. You are choosing and adapting a market offering.
Commodity work needs standards, reliability, cost control, security, automation, and operational excellence. You are not trying to express your soul through payroll, authentication, or generic compute.
This is obvious in theory and violated constantly in practice.
The common anti-pattern is prestige attachment. Teams want to keep owning work after it has commoditized because ownership used to confer status. Leaders call it strategic because it is familiar, not because it still differentiates the company. Engineers rebuild tools the market already solved. Operators defend bespoke processes because they understand them. Finance cuts experimental work because its metrics look worse than mature work.
Evolution also creates emotional whiplash. The capability that made a team special can become table stakes. That does not mean the team failed. It means the world moved.
AI will accelerate this discomfort. Many capabilities that feel magical today will become expected. Summarization, classification, drafting, extraction, and simple agentic workflows are already moving down the maturity curve in many contexts. The value will shift toward proprietary context, workflow integration, trust, distribution, and the ability to change operations around the technology.
Commodity does not mean unimportant. Electricity is commodity and civilization notices when it fails. Identity is commodity and a breach can destroy trust. Model access may commoditize while still being operationally critical. The question is not whether a capability matters. The question is how it should be managed at its current stage.
That distinction saves money and attention.
For operators, the evolution axis is a management selector. It tells you how to staff, budget, govern, and review the work. Do not ask a genesis team for five-year ROI precision. Do not let a commodity team run indefinitely on artisanal heroics. Do not evaluate a custom workflow only through vendor feature checklists. Do not let product-like tools sprawl without ownership.
The field test: list the ten capabilities that matter most to one strategic initiative. For each, answer two questions.
- What stage do we claim it is in?
- How are we actually managing it?
The gap is where waste lives.
This is part 5 of 10 in Wardley Mapping for Operators.
