When AI Is Widely Available, Where Does Competitive Advantage Come From?

October 17, 2026 6 min read

Imagine that your closest competitor buys the same AI tools your team uses. It can draft proposals, summarize documents, and search internal information with similar technology.

What would still give customers a reason to choose your business?

I find that question more useful than asking whether AI itself is a competitive advantage. Access can help a business improve, but those improvements face a competitive test: how much of the benefit survives when others can do something similar?

My view is that the stronger opportunity lies in connecting AI with a clear customer need, relevant knowledge, and an organization capable of acting on what it learns. Whether that becomes a lasting advantage depends on how difficult the resulting capability is to match.

Productivity gains face a competitive test

An improvement can be worth making even when competitors can copy it. Less administrative work, quicker responses, or fewer mistakes can improve a business’s economics. Some investments may be necessary simply to meet customer expectations.

The financial outcome still depends on competition. If several providers can deliver comparable work at lower cost, customers may gain bargaining power and prices may fall. A business could become more productive while retaining little additional profit.

Hamilton Helmer’s 7 Powers offers a useful distinction between an economic benefit and the barrier that helps sustain it against competition.¹ Lower costs can improve profitability. The strategic question is what makes it difficult for competitors to match or undermine that cost advantage.

For an AI investment, I would ask two questions together: what improves, and what makes that improvement difficult to match?

The same tool can support different businesses

Consider two hypothetical maintenance companies serving industrial customers.

The first uses AI to turn technicians’ notes into polished service reports. Staff review the drafts, make corrections, and send them to customers. If this reduces administrative effort without weakening accuracy, it is a useful improvement.

The second uses the same technology within a broader approach to serving customers whose operations depend heavily on equipment reliability. It has maintained consistent records of equipment, recurring faults, service conditions, and completed work. Technicians can retrieve relevant history, review AI-generated summaries, and decide whether a recurring issue warrants further investigation.

The company also follows up. Did the recommended work resolve the problem? Did the fault return? What did the technician miss? Verified findings become available for future service decisions.

These are different uses of organizational knowledge. The second company is trying to make accumulated experience useful across its team and over time. Its proposed customer value is more reliable maintenance advice grounded in the customer’s equipment history.

That proposition still needs evidence. Better records and more software do not establish that customers experience fewer disruptions or will pay for the service. Review time, data maintenance, and follow-up also have costs.

In Power and Prediction, Ajay Agrawal, Joshua Gans, and Avi Goldfarb describe the potential of redesigning systems of decisions around AI, including the human judgement and complementary changes those systems require.² Applied to this example, that means examining how information leads to a service decision, action, and feedback.

A faster report may be enough for the first company’s objectives. The second company’s larger commitment makes sense only if the resulting service creates enough additional value to justify it.

Knowledge becomes valuable through use and feedback

Company-specific information can contribute to differentiation when it improves a consequential decision and competing providers cannot readily obtain an equivalent basis for that decision.

An archive of inconsistent notes may offer little help. A smaller collection of accurate records, linked to decisions and outcomes, could be far more useful.

For the maintenance company, the relevant knowledge might include how particular equipment behaves under a customer’s operating conditions, which interventions have worked, and when a technician should question an apparent pattern. Some of that knowledge can be recorded. Some develops through experience and discussion.

Competing in the Age of AI distinguishes learning effects from network effects: improving through data is a different mechanism from making a service more valuable as additional users join.³ That distinction matters for a small business. Accumulating records does not automatically create either effect.

Learning requires a deliberate connection between experience and changed behavior. Someone must assess the feedback and decide whether to update guidance, change a workflow, improve the information available, or adjust the AI system. Using an AI tool more often does not necessarily train it on the business’s experience.

Data access also needs a clear commercial basis. What information can the business use, what do its customer agreements permit, and can the relevant knowledge remain accessible if it changes software providers?

The strategic resource may be the ability to keep producing reliable insight from ongoing work. That ability depends on customer relationships, staff expertise, information quality, and follow-through.

Durability comes from what others must build

Even a well-designed service can be copied.

A competitor might adopt similar processes, recruit experienced employees, or buy software that makes previously difficult work easier. Technology can weaken an advantage as well as support one.

I would therefore be cautious about calling every customized workflow a defensible advantage. Helmer’s discussion of Process Power sets a demanding standard: valuable organizational practices that require an extended commitment to replicate.¹ Documentation alone does not meet it.

For the maintenance company, a stronger position might develop through years of relevant service history, experienced technicians, established customer access, and routines that consistently turn findings into better decisions. A competitor would need to assemble a comparable combination and demonstrate that it works.

That is a possibility to investigate, not an automatic result of adopting AI. Customers could find an adequate alternative, a software vendor could provide much of the capability, or the business could fail to retain the people and relationships on which it depends.

The practical test is specific: what would a capable rival need to build, acquire, or learn to match the result, and how long might that take?

Choose the capability worth building

A small business cannot pursue every AI opportunity. Preparing information, developing staff, changing responsibilities, and reviewing results all compete for limited time and money.

Those commitments should follow a choice about customers and value. A company competing on fast, straightforward service may need a different system from one serving customers with complex reliability requirements. Adding more analysis to every job could damage the first company’s economics while helping the second.

The investment decision should make four things clear:

Customer value: What important outcome should improve, and how will the business know?
Economics: What remains after software, implementation, review, and ongoing maintenance costs?
Competitive position: Which parts could rivals readily match, and which might require sustained effort?
Organizational learning: What useful knowledge or capability will remain as the business completes more work?

Sometimes the right answer will be a modest productivity improvement using standard software. Sometimes it will be a deeper change to the service model. The scale of the commitment should reflect the strength of the case.

When AI is widely available, competitive advantage can still come from distinctive customer access, expertise, accumulated knowledge, and activities that work well together. AI may strengthen those assets or make some of them easier to replace. Each business needs to examine which is happening.

Before the next investment, return to the competitor with the same tools.

What would it still need to build to match what your customers value?

References

Helmer, Hamilton. 7 Powers: The Foundations of Business Strategy. Deep Strategy, 2016. Introduction, benefit and barrier distinction; Chapter 7, “Process Power.” Author’s book website.
Agrawal, Ajay, Joshua Gans, and Avi Goldfarb. Power and Prediction: The Disruptive Economics of Artificial Intelligence. Harvard Business Review Press, 2022. Chapter 8, “The System Mindset,” and Chapter 12, “Accumulating Power.” Publisher information.
Iansiti, Marco, and Karim R. Lakhani. Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World. Harvard Business Review Press, 2020. Chapter 6, “Strategy for a New Age.” Publisher information.