As AI adoption grows in your organisation, the number of AI use-case ideas will multiply rapidly. Without a structured intake process, your company faces two problems:
An AI use-case intake process creates a fair, transparent system for capturing AI ideas from anywhere in the organisation and routing them through evaluation, prioritisation, and (if approved) implementation.
Any employee can submit an AI use-case idea through a standard intake form.
The AI governance committee or designated reviewer conducts a quick assessment to filter out duplicates, out-of-scope requests, and clearly infeasible ideas.
Promising use cases are scored against standardised criteria covering business value, feasibility, risk, and alignment.
Scored use cases are ranked and placed on the AI project backlog.
The top-priority use cases are approved for implementation and assigned to an AI champion or project team.
The use case is implemented, measured, and reviewed. Learnings feed back into the process.
| Field | Entry |
|---|---|
| Name | |
| Department | |
| Role | |
| Date | |
What is the current process or task? [Describe the current way this work is done, without AI]
What problem does this solve or what opportunity does it create? [Describe the pain point, inefficiency, or missed opportunity]
How would AI improve this process? [Describe specifically how AI would be used — which tool, what inputs, what outputs]
Who would benefit? [List the team, department, or stakeholders who would benefit]
How often is this task performed?
Estimated time currently spent on this task: [Hours per week or per occurrence]
What data would be used as input to the AI tool? [Describe the data types involved]
Does this data include personal data?
Does this data include confidential or client data?
What is the impact if the AI output is incorrect?
Estimated time saved per week/month: [Hours]
Other expected benefits: [E.g. improved quality, faster turnaround, better customer experience, reduced errors]
Is this a quick win (can be implemented in 1-2 weeks) or a strategic initiative (requires 1-3 months)?
Use these criteria to score each submitted use case on a 1-5 scale:
| Score | Criteria |
|---|---|
| 5 | Major time/cost savings, directly impacts revenue or customer satisfaction |
| 4 | Significant productivity improvement for a large team |
| 3 | Moderate improvement for a department |
| 2 | Minor convenience improvement |
| 1 | Nice to have, minimal measurable impact |
| Score | Criteria |
|---|---|
| 5 | Can be done immediately with existing approved tools, no custom development |
| 4 | Requires minor configuration or workflow adjustment |
| 3 | Requires new tool approval or moderate setup effort |
| 2 | Requires custom development or significant integration work |
| 1 | Technically very challenging, uncertain feasibility |
| Score | Criteria |
|---|---|
| 5 | No personal data, low impact if output is incorrect, no regulatory concern |
| 4 | Minimal personal data, low to medium impact, standard compliance |
| 3 | Some personal data or medium impact, requires human review |
| 2 | Significant personal data or high impact, requires careful governance |
| 1 | Critical data or impact, major regulatory considerations |
| Score | Criteria |
|---|---|
| 5 | Directly supports company strategic priorities and AI roadmap |
| 4 | Supports departmental goals and demonstrates AI value |
| 3 | Moderately aligned with company direction |
| 2 | Tangentially related |
| 1 | Does not align with current priorities |
Composite Score = (Business Value × 0.30) + (Feasibility × 0.25) + (Risk Level × 0.25) + (Alignment × 0.20)
Score ranges:
The AI governance committee or designated reviewer:
For use cases that pass triage:
The AI governance committee:
| Field | Details |
|---|---|
| Use case ID | [AUTO-GENERATED] |
| Status | Submitted / In Triage / In Evaluation / Approved / In Progress / Completed / Deferred / Rejected |
| Assigned to | [CHAMPION OR TEAM] |
| Start date | [DATE] |
| Target completion | [DATE] |
| Actual completion | [DATE] |
| Results | [MEASURED OUTCOMES] |
The intake process only works if employees actually use it. To encourage submissions:
The most common failure mode for AI use case intake is excessive bureaucracy that discourages submissions. When employees must complete ten-page business case documents before an AI idea receives initial review, only the most persistent champions submit proposals while valuable grassroots ideas from frontline workers never reach evaluation. Effective intake processes use lightweight initial submissions — a one-page form capturing the business problem, estimated impact, and data availability — with detailed business case development reserved for ideas that pass initial screening.
Centralized intake models funnel all AI proposals through a single governance committee. This ensures consistent evaluation criteria and prevents duplicate investments but creates bottlenecks when submission volume exceeds committee capacity. Distributed models delegate initial screening to departmental technology leads who forward vetted proposals to a central committee for cross-organizational prioritization. Hybrid models increasingly represent best practice: departmental leads conduct feasibility triage using standardized criteria, while the central committee handles strategic prioritization, resource allocation, and governance approval for proposals that pass departmental screening.
Organizations progress through three maturity stages in their AI use case intake processes. Stage one (reactive): individual departments purchase AI tools independently without centralized awareness, creating shadow AI risks and duplicate investments. Stage two (controlled): a centralized intake process captures proposals, applies consistent evaluation criteria, and coordinates resource allocation across competing priorities. Stage three (strategic): the intake process evolves into a continuous innovation pipeline where proactive scanning identifies high-value AI opportunities before departmental submissions, the AI center of excellence mentors submitters to strengthen proposals before formal review, and portfolio-level optimization balances quick wins against transformational investments based on organizational capacity.
Organizations should publish an internal AI use case catalog documenting approved and deployed use cases across all departments. This catalog serves dual purposes: demonstrating organizational AI maturity to incoming proposals and inspiring employees in departments that have not yet identified AI opportunities by showcasing successful implementations in peer departments. Catalogs should include implementation timelines, resource requirements, and quantified outcomes for each documented use case.
Mature intake processes should incorporate a technical feasibility pre-screening stage using standardized checklists before proposals reach the evaluation committee. Pre-screening criteria include: data availability verification through warehouse inventory audits, API compatibility confirmation with existing middleware orchestration layers like MuleSoft or Workato, estimated compute requirements mapped against provisioned cloud GPU quotas, and preliminary vendor shortlisting comparing SaaS options against open-source alternatives hosted on internal Kubernetes clusters.
Effective AI use case intake processes complete initial screening within two weeks of submission and full evaluation within six weeks. The initial screening phase should assess basic feasibility: is the proposed use case technically achievable, does it align with organizational AI strategy, and are the required data assets available? This screening should take no more than five business days. Proposals passing initial screening enter detailed evaluation covering ROI projection, resource requirements, risk assessment, and governance review. This phase should complete within four additional weeks. Organizations that allow intake processes to extend beyond six weeks lose submitter engagement and signal that AI innovation is not a genuine organizational priority.
An effective intake form balances comprehensiveness with submitter-friendliness by capturing essential information in a single page. Required fields should include the business problem statement in plain language without assuming AI knowledge, the current process and its pain points quantified where possible (hours spent, error rates, customer complaints), the proposed AI solution at a conceptual level, the expected business impact with rough estimates of time savings or revenue improvement, data availability indicating what relevant data already exists and in what systems, and the sponsor name identifying which manager supports the proposal. Optional fields can include competitive examples showing how other companies have addressed similar problems and implementation timeline preferences. Avoid requiring detailed technical specifications, formal ROI calculations, or vendor shortlists at the intake stage.
