← All insights
AI strategy

AI business models beyond cost cutting: where new revenue actually appears

The first AI business case is usually efficiency. The larger opportunity is redesigning what can be sold, how it is delivered and which data compounds over time.

Most AI presentations start with savings: fewer manual operations, faster support and lower production cost. Useful, but incomplete. The larger opportunity is to create a product, revenue stream or market that could not exist with the previous cost and speed structure.

The 2025 McKinsey global survey illustrates the gap. Eighty-eight percent of respondents said their organizations regularly used AI in at least one function, yet most were still experimenting or piloting. Only 39 percent reported enterprise-level EBIT impact. Adoption is broad; transformation is not.

Efficiency is the entry point

Automation is measurable and easier to approve. A company reduces handling time, improves code delivery or supports more customers with the same team. The danger is that every competitor can buy a similar efficiency tool. Cost advantage erodes when the capability becomes standard.

New revenue starts with a changed constraint

AI makes some activities cheaper, faster or individually configurable. Ask what becomes sellable after that change:

  • A service that was too expensive for smaller customers.
  • A product personalized for each account.
  • A workflow delivered continuously rather than as a quarterly project.
  • A result priced by usage or outcome instead of access.
  • A data product created as a by-product of operations.

Five patterns worth testing

1. Expert workflow as software

AI can package part of an analyst, planner or operator workflow into a repeatable product. The value is not generic chat; it is domain data, process design, verification and integration.

2. Personalized production at scale

Content, recommendations, training and interfaces can adapt to the account while sharing one production system. The business model may combine a platform fee with usage.

3. Outcome delivery

If the system completes a workflow rather than merely providing a tool, the price can move closer to the verified business result. This requires measurement and clear control boundaries.

4. Data network effects

A product improves as it sees more legitimate, consented and relevant operational data. The moat is created by the feedback loop, not by calling an external model.

5. Human-and-agent operations

A service can combine AI speed with human responsibility for exceptions. This hybrid may reach markets where full automation is still unsafe or untrusted.

Questions for a founder

  1. Which customer constraint changes because of AI?
  2. What new willingness to pay appears?
  3. Which part is proprietary: data, workflow, distribution or trust?
  4. How do variable model costs affect margin?
  5. Where must a human remain accountable?
  6. What gets stronger after the thousandth transaction?
AI strategy takeaway: do not measure only how many hours AI removes. Measure how much new value the company can deliver and capture because the old constraint has disappeared.

Updated data: McKinsey, The State of AI 2025.

Adapted from an original Telegram post Telegram · 22 May 2025 →
Connect

Have a product or business challenge to discuss?

I work with founders and teams on product strategy, monetization and sustainable growth.

Schedule a call