AI Is Not a Strategy

Hiring artificial intelligence for the right job, and knowing when the right job is no job at all.

There is a growing pattern across organizations implementing artificial intelligence. The technology is impressive. The demonstrations are persuasive. The executive mandate is often clear. Vendors can show automation, summarization, content generation, workflow orchestration, predictive analytics, and conversational interfaces. The surface area of possible use cases appears almost unlimited.

And yet many AI initiatives are failing to produce durable business value. The failure is not always technical. It is not always caused by poor data, weak integration, model hallucination, employee resistance, or immature change management. Those issues matter, but they frequently sit on top of a deeper strategic problem.

Many companies are beginning with the wrong question. They ask:

That question sounds practical, but it is incomplete. It treats AI as a capability searching for a home. A better question is:

This distinction matters because AI is not strategically neutral. It changes how work is performed, how customers experience the business, how employees make decisions, and how value is created or removed. If AI is deployed against the wrong job, or against the wrong part of the value chain, it can weaken the very advantage the company depends on.

Clayton Christensen’s Jobs-to-be-Done framework provides a useful starting point. Customers do not merely buy products and services. They hire them to make progress in a specific circumstance. The same logic applies inside the enterprise.

A customer may be trying to resolve a billing issue late at night. A manager may be trying to understand why gross margin changed. A compliance officer may be trying to review contracts for risk. A sales leader may be trying to determine which accounts require attention. A finance team may be trying to close the books with fewer errors and less rework. These are jobs, and they carry functional, emotional, and social dimensions.

AI implementation should begin with this understanding. Not with the tool. Not with the vendor. Not with a generalized desire to “do something with AI.” The first discipline is to understand the job and the circumstance in which it arises.

“We need an AI chatbot” is not a strategy; it is a proposed mechanism. The better formulation is: “Our customer needs to resolve a billing discrepancy at 11:00 p.m. without waiting on hold.” That framing clarifies the circumstance, the desired progress, and the outcome that matters, and it prevents the company from confusing the technology with the job itself.

A company can correctly identify the job and still deploy AI in a way that damages the business. This is where Michael Porter’s work on competitive advantage becomes essential. Porter’s lens asks a different question: How does this company actually win?

Does it compete through cost leadership, where customers value low price, speed, consistency, and efficiency? Through differentiation, where customers value service, expertise, quality, trust, customization, or status? Or through focus, serving a narrow market with specialized knowledge or a tailored operating model?

These distinctions are critical, because the same AI use case can be strategically correct in one company and strategically destructive in another. A self-service AI support agent may strengthen a cost-leadership business if customers value speed and low price, and weaken a differentiated service business if customers value relationship, judgment, and human responsiveness. The issue is not whether AI can perform the task. It is whether AI should perform that task in that business model.

The most dangerous AI failures occur when companies hire AI for a job that conflicts with their source of customer value. This is especially visible in customer service.

For some companies, customer service is primarily a cost and throughput function. Customers want fast answers, simple transactions, low prices, and minimal friction. In that environment, AI-enabled self-service can be strategically appropriate, reducing wait times, improving consistency, and reinforcing the firm’s cost position.

For other companies, customer service is not merely support. It is part of the product. Customers stay because they trust the people. They pay more because they value judgment, continuity, and specialized advice. They are not buying a transaction; they are buying confidence. Using AI primarily to remove human interaction may reduce cost while damaging differentiation.

That is not transformation. It is strategic erosion, and failing to recognize the distinction is one of the central mistakes in AI strategy.

A useful executive question is deceptively simple: What do our customers love about us? It may sound informal, but it is strategically precise. Whatever the answer, AI should be evaluated against it.

The test is not whether AI is impressive. The test is whether AI amplifies the source of customer preference.

AI initiatives generally serve one of two strategic purposes. They either lower the floor or raise the ceiling.

The problem occurs when leadership fails to distinguish between the two. A company may present an AI initiative as innovation or customer-experience improvement when the underlying objective is cost reduction. Cost reduction is not inherently wrong, but if the cost being reduced is tied to customer value, the initiative requires much deeper scrutiny.

In a differentiated business, AI may create more value behind the scenes than at the customer interface. It can summarize prior interactions, surface service risks, prepare account history, recommend next actions, and help humans deliver better service without replacing the relationship. There, AI is not hired to remove the differentiator. It is hired to strengthen it.

Before approving an AI initiative, leadership teams should be able to answer seven questions.

Many organizations measure AI progress through activity metrics: users trained, prompts submitted, employees with access, responses generated, use cases launched. These may indicate motion, but they do not prove strategic value.

The central question: Did AI help the job get done better? If the answer is no, then adoption does not matter.

AI initiatives should not be managed as a loose collection of experiments. They should be classified by strategic intent — cost, customer experience, employee leverage, decision quality, risk reduction, or product differentiation.

Without this classification, AI portfolios become internally contradictory. One team uses AI to personalize experience while another removes the human interaction that made personalization credible. One function raises quality while another increases throughput at its expense. One department optimizes for cost while another depends on the premium service those costs support. This is not an AI problem. It is a strategy problem.

The organizations that succeed with AI will not necessarily be those with the most use cases. They will be those with the clearest logic for where AI belongs in the business model: knowing what to automate, what to augment, what to govern, what to protect, and what should remain human because it is central to the company’s value.

AI strategy should not begin with “Where can we use AI?” It should begin with a clearer, more disciplined question: What job are we hiring AI to do, and does that job strengthen the reason customers choose us?

Jobs-to-be-Done identifies the progress a customer, employee, or function is trying to make. Competitive strategy determines whether assigning AI to that job reinforces or undermines the company’s position. That synthesis is essential.

AI can lower cost, raise value, reduce risk, improve judgment, increase consistency, and strengthen service. It can also commoditize a differentiated business, weaken customer trust, and erode competitive advantage. The difference is not the technology alone. AI fails when it is hired for a job the business strategy never should have assigned to it.