AI, in your work.

From billable hours to outcomes

The official Directions listing places Sharka Chobot of Neural Impact's session on value and pricing in the AI era on October 29, 2026. Its summary discusses pressure on billable hours and a shift toward outcome-based models, alongside uncertainty over consumption costs.

Nexo's editorial interpretation is that this conversation needs an initial test: demonstrate what improves for the customer when a task requires less work. The method below is our own proposal for designing that test, rather than a reproduction of the slides or a result attributed to the speaker.

Three measures to keep separate

We propose keeping three separate accounts when evaluating the pilot. Available time can be reassigned to other tasks, but converting it into money using an internal hourly rate does not, by itself, demonstrate that a payment has fallen.

There may also be value in handling more transactions, reducing delays or improving quality. In that case, measure the outcome directly and avoid counting the same benefit twice: as freed time and as a separate financial saving.

Agree on the test before starting

The pilot should begin with a specific operation and a shared definition of completed work. Counting generated responses may be insufficient when the customer needs resolved and accepted cases.

We propose recording these conditions in a short plan agreed by the people responsible for the process and its evaluation.

• Baseline: volume, case difficulty, total time and quality in the current process. Include exceptions and work that currently requires manual correction.
• Intended outcome: agree on the metric, improvement threshold, minimum quality, budget and observation period before seeing results.
• Evidence: record each case, human review, errors, retries, consumption and time to acceptance. Separate setup, training and recurring maintenance.
• Attribution: compare equivalent cases and, where feasible, retain a reference group or period without AI. Record simultaneous changes in staffing, volume, data or procedures.
• Responsibility: establish who validates quality, who checks costs and who decides whether to scale, adjust or stop the test.

Fictional example: reviewing inventory issues

Imagine a team preparing proposed resolutions for inventory issues. During the test, an assistant generates drafts and a person reviews every case. All figures below are illustrative assumptions created by Nexo; they do not describe a customer, a product or an experience reported in the session.

We compare two four-week periods with 500 cases of equivalent difficulty, accepted under the same quality criteria. The assumed recurring balance would be 30 hours freed and €200 in additional expenditure. If the same team also provides the 12 initial hours of setup and training, the first period would leave 18 net hours.

Because payroll remains unchanged and no other payment disappears in this scenario, realized savings are zero. Justifying the additional expenditure would require evidence of a useful outcome from reassigning those hours. The example does not establish a positive financial return.

Decide before negotiating the split

Closing the pilot should produce a decision supported by records. A favorable result in a small sample also does not demonstrate that the same performance will hold with more users, harder cases or greater volume.

In Nexo's view, the subsequent negotiation should distinguish the service price, variable usage costs and any outcome-linked compensation. If benefit sharing is proposed, both parties need to agree on which benefit counts, how it is attributed, which costs are deducted and how long it is measured. Freed time should become a financial basis only when both parties understand and accept its valuation.

• Scale: the improvement repeats, meets quality and cost thresholds, and there is a verifiable use for the freed capacity.
• Adjust: benefits emerge, but review, errors, consumption or training prevent the pilot from meeting its objective. Change one condition and measure again.
• Stop: the improvement disappears when all costs are included, fails the agreed quality standard or cannot be attributed on the available evidence.

Scope and availability

This design is an evaluation proposal and does not assume that a feature is available in Business Central or any other product. Before running it, the chosen solution's current documentation would need to establish access requirements, licenses, supported regions and languages, usage limits and pricing.

The session listing establishes the commercial discussion, but it does not replace those technical checks or demonstrate savings in a specific organization. For this pilot, the proposed success criterion is an accepted outcome with measured quality and cost, whose contribution can be explained.