Skip to main content

Use case document

Purpose​

The use case document is a structured description of an initiative: which process is changing, which problem is being solved, which AI product fits, which data is needed, and which impact is expected.

It is needed at the assessment stage in order to make a decision: admit the initiative to delivery, return it for clarification, merge it with a similar initiative, or reject it.

Document structure​

Current process​

  • Description of the process where AI is planned to be deployed.
  • Process participants and their roles.
  • Bottlenecks, inefficiencies, manual operations.
  • Process frequency and operation volume.

Business problem​

  • A clear statement of the problem the initiative solves.
  • Scale of the problem: frequency, volume, impact on the business result.
  • Why the problem is not solved by current means.
  • What will happen if the problem is not solved.

Key metrics and current values​

  • A list of metrics affected by the initiative (for example, application processing time, error rate, conversion).
  • Current metric values with the source and measurement period indicated.
  • Who owns these metrics.

Proposed solution​

  • Type of solution: classification, generation, recommendation, automation, information extraction.
  • A brief description of the approach (model, algorithm, architecture — at the conceptual level).
  • The solution's place in the process: which steps are automated or augmented.
  • Where the human stays in the loop.

Expected impact on metrics​

  • Target metric values after deployment.
  • Justification of the expectations: analogues, expert assessment, results of past initiatives.
  • Date or period of the impact check.

Data availability assessment​

  • Which data is needed to train and run the model.
  • Data availability: yes / partially / no.
  • Data quality: completeness, freshness, labeling.
  • Constraints: personal data, confidentiality, regulatory requirements.
  • Data owner.
  • Expected processing perimeter.

Impact hypothesis​

  • Statement: "If we deploy [solution], then [metric] will change by [amount], which will lead to [business impact]".
  • Type of impact: cost reduction, revenue growth, productivity increase, risk reduction.
  • Preliminary estimate of the economic impact (annual).
  • Confidence level: high / medium / low.

Participants​

  • Business owner of the process.
  • Users of the solution.
  • AI office.
  • Product owner.
  • Data owner.
  • Security and architecture if needed.

Target AI product​

  • Which AI product the initiative belongs to.
  • Whether existing components are reused.
  • Whether there is a suitable product delivery track.
  • What to do if there is no product.

Scalability assessment​

  • The potential to replicate the solution to other processes, units, segments.
  • Scaling constraints.
  • The possibility of reusing the result as a product component.

Use in the process​

The use case document is prepared by the initiator together with the business owner and the AI office. The AI assistant can assemble a draft from a free-form description, but responsibility for the content remains with the owners.

Decisions following the review:

  • admit to delivery;
  • return for clarification;
  • choose a different product;
  • merge with a similar initiative;
  • postpone;
  • reject.

Details of the verification process are in validate AI use case.