From AI Capacity to New Customer Value

Compare the customer promises before choosing the technology

The maker should first examine a less demanding investment: improving the support it already provides. Searchable guidance, clearer intake and better routing might resolve the supervisor’s problem without creating a subscription obligation. The new offer earns attention only if preparing a usable plan supplies a difference customers value.

The proposal develops Know-how as a service in Possibilities. AI Strategy introduces growth opportunities; this test examines one customer promise. The portfolio choices in Fund the Operating Model Behind AI Value remain separate: a promising offer must still compete for funding and scarce people. From Idea to Operating Case supplies the discipline for tracing an approved change into accepted outcomes and realized value.

Separate a usable brief from value for the customer

Limit the offer to one supported machine family and verified machine identifiers, usage and maintenance history. The brief identifies applicable routine tasks, preparation and questions requiring the customer’s technician. It excludes emergency diagnosis, remote control, repair authorization, safety certification and an uptime guarantee.

The maker identifies missing information within one business day. Once intake is complete, it promises the reviewed brief within two business days and answers the included clarifications within one business day. Unsupported requests go to an identified specialist route. Incomplete intake and waiting time remain visible when judging the customer experience.

An accepted planning brief must match the machines, use current permitted guidance and pass qualified review. The customer’s maintenance lead confirms that it is usable for planning. The Outcome owner records acceptance after ten business days of follow-up without an unresolved material error. A correction does not create a second accepted outcome.

Customer value needs separate evidence. Did the supervisor use the brief to schedule eligible work? Did total planning effort fall after including input preparation, checking and clarification? Did the customer choose to renew at the stated price? A technically acceptable brief can remain unused or add work. The experiment tests these questions directly; it does not establish improved uptime or maintenance performance.

The equipment maker remains responsible for investigating errors, correcting briefs and notifying affected customers. The Operating owner provides qualified backup, or communicates the interruption and arranges a credit or refund when the promise cannot be met. The customer retains maintenance authorization and safety decisions. That retained authority does not excuse an inaccurate planning service.

Use assistance where it changes preparation effort

Rules validate eligibility, identify missing fields and perform approved calculations. AI assistance assembles relevant guidance from permitted sources and varied service notes, then drafts the brief. Qualified review connects that preparation to a deliverable promise, as the workflow shows. This defined sequence does not require an agent to choose its own next steps.

The allocation of authority stays explicit. Recommend covers the draft; Act permits the controlled workflow to deliver the approved version. Escalate routes conflicting guidance, unsupported equipment or uncertain applicability to the designated specialist and keeps the brief held while it waits. Never Delegate covers maintenance authorization, consequential safety judgments and changes to the offer’s scope.

Compare that design with indexed guidance, templates and human preparation on matched cases. NIST’s AI RMF Playbook includes consideration of non-AI alternatives when establishing the context for a proposed system. [1] If simpler preparation can meet the same promise at lower full cost, use it.

The maker must prepare a controlled knowledge set before selling access to its expertise. The Knowledge owner maintains approved planning guidance, machine applicability and source revisions. Permission to retrieve an internal note does not authorize disclosing it to a customer. Customer histories stay segregated, with agreed retention and permitted use. A source correction or rights change must identify affected briefs and reach their recipients. NIST also addresses risks involving third-party information and intellectual property rights. [1]

The service-business leader is the Outcome owner for the promise, acceptance criteria and funding. The support manager is the Operating owner, with named specialists and a qualified deputy who can approve or withhold delivery. The Technology owner maintains access and delivery controls; the Control lead coordinates control evidence. Finance challenges the economics, and the Sponsor resolves resource conflicts. These duties require assigned capacity. If qualified cover is unavailable, the Operating owner holds new intake rather than drawing silently on existing customer commitments.

Expose the change in delivery economics

Assume conventional preparation of the proposed brief requires two specialist hours to gather and apply guidance, followed by one hour of judgment and checking. The AI-assisted design targets 15 minutes of preparation and 45 minutes of specialist review. Both estimates exclude separately budgeted support and knowledge maintenance; neither is an observed saving.

At an assumed fully loaded resource cost of $100 per specialist hour, the ten-customer preparation requirement would fall from 30 hours to ten. The Operating owner and the managers controlling those people must agree the assignments and confirm residual capacity for existing services. Lower required effort does not itself release usable hours or reduce payroll.

The table compares the same new service at a $300 price per customer per four-week cycle. It then shows eight customers renewing. Specialist capacity and support are reserved for ten customers through both cycles, so those commitments remain funded when two leave. Technology usage alone varies with active subscriptions in this illustration.

Contribution here means revenue less all defined recurring resource costs. It is neither cash savings nor profit after initial investment and acquisition. Reserved support includes customer questions, correction, recovery and remaining follow-up for departing customers. The shared allocation covers the offer’s reservation of billing, monitoring and controlled delivery services. It stays fixed for the agreed period. Charge that allocation once to the offering and reconcile it to the central total; do not add both views together.

Retained volume must carry the commitments that remain. Calling the two unused specialist hours a saving would conceal the reservation. A later manager-approved reassignment might change the next period’s resource allocation; it cannot erase this period’s cost or be credited to two uses.

If all ten first-cycle briefs complete acceptance within the cycle, Cost per accepted outcome is $2,100 ÷ 10 = $210. At eight renewals with eight accepted briefs, it is $2,060 ÷ 8 = $257.50. This measure is total recurring cost of running the capability in a period divided by outcomes accepted by the owner in that period. Include failed work and correction in cost, count each outcome once, and disclose any acceptance or cost carried into another period. With zero accepted outcomes, it is not calculable. Renewal and customer effort remain separate measures.

Keep initial investment and customer acquisition visible outside that recurring calculation. Assume $5,000 for initial design, discovery and setup, plus $100 per initial customer for acquisition and onboarding: $1,000 for ten. These are different costs from the $200 recurring shared-service allocation. Two cycles with ten then eight customers would contribute $1,240 before those $6,000 of costs and any additional follow-up-period operation. The experiment has not recovered its investment.

For expansion, identify future avoidable costs, new acquisition and onboarding, additional setup and the alternative use of people. Do not charge already incurred investment again as a new expense, or hide it when reporting the experiment’s cumulative result. Any additional closure, credits or refunds must enter the actual record. One renewal cannot establish a long customer lifetime or justify a speculative lifetime-value forecast.

Test customer behavior alongside the service

Begin with 20 qualified small operators in the same machine family who have an upcoming planning need. Examine their current workaround and comparable planning records, show the improved-guidance alternative and ask them to choose. NSF’s I-Corps program uses customer discovery to investigate potential product-market fit and inform a business model. [2] Here, conversations shape the offer; paid use and renewal provide the subsequent behavioral evidence.

Limit the experiment to ten customers and twelve weeks: two for preparation, two four-week service cycles and two for final follow-up. Cap total resource cost at $12,000, including internal effort. Within that limit, the $5,000 setup allowance and $1,000 acquisition/onboarding allowance leave up to $6,000 for recurring delivery, follow-up and recovery. Revenue does not enlarge the spending authority.

Set the decision thresholds before enrollment. Ten customers must pay $300 for the first cycle; at least eight must use the brief to schedule eligible work and report at least 30 minutes less total planning effort than in a comparable prior cycle; eight must renew at the same price. The 30-minute reduction is one indicator within the broader proposition, alongside actual use and renewal. Record customer effort from input preparation through checking and clarification, and examine case differences. These are illustrative commercial thresholds, not statistical proof of a market.

Separately, at least 90% of eligible briefs must meet the two-business-day delivery commitment, every issued brief must receive qualified approval, and accepted briefs must complete follow-up. Count eligible failures and difficult cases. Require at least 20% contribution after the defined recurring costs at actual retained volume, net of credits and refunds. Existing contractual commitments and protected staffing must remain intact, with no unplanned overtime caused by the trial.

Monitor corrections, customer complaints and source changes while the experiment runs. NIST’s post-deployment guidance includes feedback, intervention, recovery and controlled change. [3] An unsafe recommendation that escapes review or an unauthorized disclosure triggers an immediate pause and investigation. Qualified coverage failures suspend new intake. These incident actions protect customers while the owner decides the future of the offer.

Let the renewal evidence change the commitment

Suppose ten customers pay, eight use the brief and meet the effort-reduction threshold, and eight renew. Assume delivery acceptance and existing-service safeguards also hold. That would support a narrow claim: some customers in this selected cohort value this offer enough to use it and buy another cycle. It would not establish broader demand, sustained retention or better equipment performance.

Plotting contribution at each customer volume makes the remaining commercial constraint visible.

Under the reserved-capacity assumptions, the retained cohort misses the commercial threshold. I would choose Redesign for the next commitment, complete the paid service obligations and withhold expansion. The maker would need a credible change in scope, frequency, price or delivery work, with customer acceptance of the revised promise tested again. It should not assume that replacement customers will immediately fill the two vacancies.

Stop is appropriate if customers do not value the offer, required rights cannot be secured, or no feasible design meets the economics. Narrow may preserve a useful equipment segment or simpler promise. Continue retains a justified bounded scope with a defined review; Expand requires evidence and funded capacity for the proposed increase. The formal choices remain Expand, Continue, Narrow, Redesign and Stop. Any future delegation of greater authority requires its own decision.