How AI agents could reshape industry value chains

United Airlines restricts unauthorised AI access

On 11 November 2026, United Airlines announced restrictions on unauthorised third-party AI agents accessing its platforms, including indirect access through travel agencies and aggregators. United cited the protection of customer data and accounts, along with the safety of booking transactions. Those concerns matter when software is acting on someone’s behalf.

There is also a wider strategic issue. Travel distribution has long involved work that customers are willing to pay someone else to do: finding options, interpreting conditions and arranging a purchase. If an AI agent can take on more of that work, the commercial value of each participant in the booking process may change. That is a strategic implication of the technology, rather than evidence of United’s motive for its decision.

Which services will still justify a fee?

An intermediary can earn its margin by reducing a customer’s search effort, bringing together fragmented supply or making a complicated transaction easier. An agent that compares suppliers and completes a purchase directly could reduce the value of those activities. Businesses whose differentiation rests mainly on making information easier to find may face particular pressure.

Other sources of value may become more important. An intermediary might negotiate access or prices that a customer cannot obtain alone, take responsibility when a trip goes wrong, or provide judgement on an unusually complex purchase. The question for management is how much customers value those services and whether the business can charge for them explicitly. Easier comparison does not remove the need for trust, accountability or help when a transaction fails.

Better comparison could put pressure on margins

Customers often compare only the most visible part of an offer. In travel, a headline fare may leave baggage, flexibility and other conditions to be checked separately. In other industries, contract terms, service levels or the cost of switching can make comparisons just as difficult. An agent with reliable access to the relevant information could assess more of the total cost against the customer’s requirements.

That could weaken margins supported by complexity or the effort involved in checking alternatives. It could also help a supplier whose better service or lower lifetime cost is difficult to explain today. Management should examine which parts of the offer would look stronger under a more complete comparison, and which depend on customers overlooking a fee, condition or alternative.

Agents could influence who reaches the customer

As customers delegate more of a purchase, an agent may become an influential point between them and a supplier. Its recommendations will depend on the information it can access, the instructions it receives and the criteria it uses to rank alternatives. A supplier may still deliver the product while having less influence over how the customer discovers and chooses it.

For a mid-market business, this creates practical questions about distribution. Can an agent accurately understand the offer? Are prices, availability and conditions accessible and reliable? Will customers still seek the business out by name, or allow software to select a supplier? The answers will affect investment in brand, direct relationships and distribution partnerships.

Assess margin exposure activity by activity

The Playbook’s industry assessment becomes more useful when management connects it to the accounts. Break revenue and service costs down by customer group and transaction type. A simple repeat purchase may be exposed to automated comparison, while an unusual requirement or a difficult exception still needs specialist support. Treating both as one revenue stream can hide where the risk sits.

For each material activity, test a plausible change in how the customer buys. If the search and comparison fee falls, can the remaining service earn enough to cover its cost? If customers arrive through an agent, does lower acquisition effort offset a weaker direct relationship or a new distribution charge? Use actual transaction records to assess those questions. A broad assumption that AI removes all intermediary value is unlikely to produce a useful business case.

• Separate fees earned from access, advice, transaction handling and ongoing support, rather than treating the intermediary’s margin as a single service.
• Identify the customer and supplier information an agent would need, who controls it and whether access could be withdrawn or become chargeable.
• Compare a narrower specialist proposition with the current offer. Test demand for the service that would remain valuable after routine search becomes easier.

Industry economics belong in the AI strategy

A useful starting point is to map the activities between a customer recognising a need and receiving the product or service. Identify who performs each activity, what the customer pays for it and where the business earns its margin. Then consider which activities an agent could perform, which require access to proprietary information, and where a person or business must remain accountable.

This is the role of the AI Strategy Assessment in The AI Playbook for Mid-Market Leaders. It examines how AI could change industry profit pools and the sources of competitive advantage. The AI Value Creation Plan then connects those choices to the capabilities and workflows the business needs to develop. Internal efficiency is part of that work. So is understanding whether the way the business earns its margin is likely to change.