A Norwegian engineering consultancy completes a familiar assignment in half the time. The customer receives the same usable report, with the same professional review. Should the price fall, should the firm keep the efficiency gain, or should it offer something different?
The answer depends on what the customer buys and what the agreement promises. My starting point is to separate three things: the cost of producing the work, the value of the delivery, and the basis on which the customer agreed to pay. AI changes the first. It does not automatically settle the other two.
Put three offers beside the same assignment
Consider a hypothetical assignment previously requiring 20 hours, sold at NOK 1,500 per hour. Revenue was NOK 30,000. At an assumed internal delivery cost of NOK 800 per hour, NOK 14,000 remained before overhead and other costs. These are illustrative figures, not market rates.
Suppose an AI-assisted process reduces total delivery time to 10 hours, including review and corrections, and adds NOK 1,000 in tool costs per assignment. Three offers now look different:
- Hourly work at the same rate: NOK 15,000 in revenue less NOK 8,000 in delivery cost and NOK 1,000 in tools leaves NOK 6,000 before overhead. The customer gains a lower bill. The firm needs sufficient demand and usable capacity to replace the lost contribution.
- A fixed fee of NOK 25,000: the same direct costs leave NOK 16,000 before overhead. The customer pays for a defined delivery and the supplier carries the risk of extra work within the agreed scope.
- A continuing service: regular updates, monitoring or access to advice may solve a recurring need. Its price needs a separate workload and demand model. A subscription is not justified merely because the supplier wants predictable revenue.
The comparison excludes VAT, selling costs, implementation, financing and tax. It is not a profit forecast. If review takes longer or the assignment is unusual, the apparent advantage can disappear. Use the guide to turning saved time into business results to test whether released hours can actually be used.
Agree on the change before billing differently
If the agreement bills time actually worked, do not present hours that were never worked as delivered hours. A fixed fee needs a clear description of the result, assumptions, exclusions, revisions and acceptance. Changing a delivery method is also a reason to check contractual requirements concerning confidentiality, subcontractors and AI use.
For an existing customer, I would bring a concrete proposal: which part becomes faster, what quality checks remain, what the customer receives, and how the price changes. A lower price can be sensible if it expands demand or strengthens a valuable relationship. Keeping the price can also be defensible when the customer receives an agreed result and the supplier still carries substantial responsibility. Neither choice should rest on hiding the production method when disclosure is required or material to the agreement.
Test what the customer values before changing the price list
Choose one repeatable service. Ask recent buyers which constraint matters: price, turnaround, certainty, access to expertise or the work they must do themselves. Do not assume that an earlier draft is valuable if the customer's approval process still takes weeks.
Offer a tightly scoped alternative to a small number of suitable customers. Record acceptance, reasons for declining, actual delivery effort, revisions and contribution after direct costs. Compare similar assignments. A discounted pilot does not establish willingness to pay the normal price.
SSB's September 2026 analysis documents growing AI use in Norwegian businesses. It does not establish that customers in your market expect lower fees. Evidence from your own offers should carry more weight in this pricing decision.
Make the commercial boundary explicit
The management team should decide who can approve a new price, how much variation is acceptable and which assignments remain unsuitable for a fixed fee. Preserve paid time for professional review. Removing the reviewers to make the spreadsheet work can undermine the service the customer is buying.
For a material change in business model, the board should challenge demand assumptions, margin sensitivity and concentration risk. It should not approve individual customer quotations. The next useful question is whether the firm is winning the right work; see how to evaluate AI in proposal preparation.
Bring one existing service to the next management meeting, with its current agreement and actual delivery costs. Compare the three offers before buying more automation. The decision is what to sell and on what terms.
Sources and scope
Sources checked on 12 October 2026. The consultancy, costs and offers are hypothetical. The pricing framework and recommendations are my analysis. Contracts and sector-specific professional requirements need their own assessment.