How to Scope an AI Service Pilot: Deliverables, Review and Acceptance Criteria

In a fictional pilot discussion, a founder asks a catalog editor, “Could you try ten AI-assisted product descriptions before we build a store integration?” The editor is interested, but wants approved specifications, editable drafts and a clear way to reject an inaccurate claim. That conversation is the beginning of a useful pilot brief.
A useful AI service pilot is a small delivery agreement. It says what you will accept as input, what you will hand over, who checks the work and when to continue, revise or stop. Keep the first delivery small enough to inspect every result.
If you are still unsure whether a buyer wants the outcome, start with the validation guide. This guide answers the next question: can you deliver one agreed result under workable conditions?
Start with a completed pilot brief
Here is a fictional example for a catalog-writing service. The buyer wants ten editable product descriptions from approved specifications. The goal is to evaluate useful drafts and review effort before investing in a store integration.
| Decision | Agreed example |
|---|---|
| Inputs | Ten approved product rows, specification sheets and brand examples |
| Deliverable | Ten editable drafts, source references and an exceptions log |
| Draft format | Agreed spreadsheet columns, 80 to 120 words per description |
| Reviewer | Catalog editor, with the ecommerce manager accepting the batch |
| Timing | Three working days after the input checklist is complete |
| Revisions | One consolidated round within the original scope |
| Review limit | Buyer agrees to track checking time against an illustrative 60-minute target |
| Cost limit | Six delivery hours and a hypothetical $300 direct-cost ceiling |
| Exclusions | Store publication, new photography, additional languages and live integration |
The quantities, timing and budget are planning choices for this example, not market standards. Replace them with limits that fit your customer and the work.
The brief is useful because it settles ordinary disagreements before they happen. A second language is additional scope. Correcting an invented specification is a defect. Those should not become the same conversation.
Decide when the work is ready to start
Use an input checklist. Confirm the product records, approved sources, permitted claims, brand examples and reviewer are available. Start the delivery clock when that package is accepted, rather than assuming missing information will arrive later.
For a product with no verified dimensions, record “dimensions not supplied” and pause that claim. Do not fill the gap with a plausible number. Ask the buyer whether to supply the missing source, omit the claim or exclude that item from the batch.
Agree on permission to process the material and how it will be stored and retained. Use approved samples while tool settings or contractual protections remain unresolved. A small pilot should not require casually sharing a whole customer database.
Give review a person and a method
The delivery owner prepares the drafts and checks them against the supplied records. The buyer's reviewer then checks usefulness, wording and suitability for the intended job. Name the person who accepts the final batch so a chain of informal comments does not become an undefined approval process.
For this catalog pilot, acceptance could require:
- Every factual statement matches an approved source.
- All ten products have a draft or an explicit blocked status.
- Drafts follow the agreed columns and word range.
- Unsupported claims are corrected before an item is accepted.
- The reviewer records correction time and the reasons for substantial rewrites.
Keep blocking defects separate from preferences. One invented measurement matters even if the other nine drafts look good. If the same failure could affect the whole batch, inspect the other items before delivering them.
A second pilot might need different checks. For meeting summaries, a reviewer could compare every assigned action with the authorized transcript and flag ambiguous speakers. The acceptance method should follow the actual deliverable.
Decide what happens when only part of the pilot passes
A small pilot can deliver useful work and still fail an important acceptance check. Agree on the treatment before delivery. Can the buyer accept individual items, or does the batch need to pass together? Who approves a replacement input? What happens to the delivery clock while the buyer supplies a missing source? Write those decisions into the brief so the first exception does not become an improvised commercial negotiation.
Return to the fictional ten-product catalog pilot. Eight drafts match their approved specifications and pass review. One draft contains an unsupported battery-life statement. Another product has no approved material specification. These are two different problems. The unsupported statement is a delivery defect to correct and review. The missing specification is an input exception that needs the buyer's decision. Neither should be quietly counted as an accepted draft.
The delivery record could say: “Eight accepted; one revision required for an unsupported claim; one blocked pending approved material specification. No store publication authorized.” The counts describe the actual state. A team can then decide whether to accept the eight usable items now under the agreement or wait for the completed batch.
Separate a correction from a paid scope change
Removing or correcting an invented product claim belongs in defect handling. Asking for a second language or a different audience introduces additional work. Record the requested change, effect on timing and cost, and the buyer's decision before doing it. Avoid using the revision allowance to excuse inaccurate work or treating every preference as free extra scope.
When a defect suggests a repeatable failure, inspect the other drafts for that same issue. In this example, check all battery and performance statements against the supplied records. Document the check and its result. A single corrected sentence does not establish that the underlying pattern is resolved.
End with a usable acceptance record: accepted items, unresolved items, correction owner, next check and the action actually authorized. That record gives both sides a clearer basis for a second order than a blanket “pilot successful.” It also reveals what software would need to support later, including partial acceptance and blocked inputs.
A completed acceptance record
| Item status | Evidence and next action | Owner |
|---|---|---|
| Eight accepted drafts | Match the approved records and agreed format; retain the checked versions | Ecommerce manager |
| One delivery defect | Remove the unsupported battery-life claim, recheck the source and submit the corrected draft | Delivery owner, then catalog editor |
| One blocked input | Ask for an approved material specification or written permission to omit that claim | Buyer |
| Publication | Still excluded from this pilot; acceptance of the drafts does not schedule or publish them | Store owner |
The acceptance record can name drafts 1 through 8 as accepted and keep drafts 9 and 10 open. If the agreement requires one complete batch, the same evidence supports a different status: batch awaiting acceptance, with eight items ready. Agree which rule applies before delivery. The numbers should describe the agreement and the actual work, rather than make the pilot look finished.
Track time before the ceiling disappears
Include intake, source preparation, generation, checking, revisions, support and exceptions. Separate initial setup from repeat delivery so you can see what a second order would really require.
In a fictional calculation, six delivery hours at an internal $40 hourly cost plus $25 in variable tools totals $265 before acquisition, overhead and taxes. Against a $300 direct-cost ceiling, that leaves $35 of headroom. One more hour would exceed the ceiling.
Agree on what happens before that point: report the cause and propose a smaller batch, revised timing, an approved paid change or a stop. Do not quietly continue and surprise the buyer with a larger bill. Track the buyer's checking time separately; shifting effort onto the customer can make a fast service unhelpful.
Hand over a package someone can use
Deliver editable drafts, their source references, unresolved exceptions and the acceptance record together. Include the agreed retention or deletion date. Make rights to customer inputs, created outputs and reusable materials clear in the agreement.
Acceptance of drafts does not authorize publishing them, sending messages or connecting the workflow to production. Keep the person responsible for that next action explicit. For the catalog example, the ecommerce manager can accept the batch while the store owner separately controls publication.
Agree on a review window and how disagreements will be resolved. Do not assume silence means acceptance. Clear terms make a small pilot easier to manage for both sides.
Price the second batch from the work you actually did
The catalog pilot has another useful result: a record of what repeat delivery requires. Before quoting a second batch, separate reusable setup from the work that returns with every order. The buyer's accepted drafts tell you what was useful; the time log tells you what it cost to produce them.
Here is a fictional six-hour breakdown: two hours preparing the initial format, one gathering and checking sources, one generating and formatting drafts, and two reviewing and correcting them. At the example internal cost of $40 per hour, those six hours cost $240. Add the assumed $25 in variable tools for the earlier $265 total.
For a comparable second batch, suppose the format can be reused but the four other hours remain. That would be $160 in labor plus $25 in variable tools, or $185 in direct cost. This is a planning estimate under explicit assumptions, not evidence that the next delivery will take four hours. New product complexity, missing specifications or a different reviewer could change it.
A useful next-order note says: “Reuse the agreed format. Quote the same ten-product scope after reviewing the new inputs. Record source checking, corrections and buyer review separately. Revisit the estimate if the input package or required output changes.” Keep the original pilot cost in your business records rather than making it disappear because the template is reusable.
Do not use the $115 gap between $185 and a hypothetical $300 selling price as take-home profit. Acquisition, overhead, taxes and other applicable costs still need treatment. More immediately, ask whether the buyer found the drafts worth reviewing. A repeatable service needs a customer who wants another accepted batch, as well as a delivery process that fits its costs.
Finish with a decision
Continue with the narrow offer when the output is accepted, review effort is tolerable and delivery fits the limits. Revise the inputs or scope when the buyer values the result but the process is too expensive. Keep it manual when expert judgment still does essential work you cannot specify reliably. Stop when permissions, inputs or recurring defects prevent a responsible delivery.
Use the accepted record as a build brief: input fields, output format, roles, exception states and cost limits. It tells a developer far more than “automate our catalog.”
If you need help organizing the original idea, view the sample playbook. If the pilot has clarified what deserves software, read the MVP hiring guide. Start your next pilot by agreeing on one deliverable and one completed worksheet.
Follow one rejected draft through correction
Suppose the catalog reviewer spots a material that never appeared in the approved inputs. Record which claim failed, return the draft for correction, and ask the same reviewer to check it against the source before accepting it. Decide whether that correction fits within the agreed time and revision limits. If the team is ready to commission a build, include this example when you compare AI MVP proposals.
Need a structure for the idea first?
The sample playbook is a planning aid, not customer evidence or a guarantee. The free diagnostic builds a personalized playbook.
Vibe Your Company Editorial Team
The Vibe Your Company Editorial Team prepares practical guides for evaluating AI-enabled businesses. This article uses hypothetical planning examples, not reported client outcomes.