# AI Pilot Scope Template

Use this template to define a focused, decision-ready AI pilot before development begins.

## 1. Business problem and owner

- Workflow name:
- Business owner:
- Decision deadline:
- Why this matters now:
- What is explicitly outside this pilot:

## 2. Current baseline

- Workflow volume per week or month:
- Current cycle time:
- Current labor cost:
- Current completion rate:
- Current error rate:
- Current escalation rate:
- Source and measurement period for these numbers:

## 3. Pilot outcome

- Verifiable end state:
- Who verifies completion:
- Where human approval is required:
- Safe state when the system cannot complete the task:

## 4. Data and integration boundary

- Approved data sources:
- Systems the pilot may read:
- Systems the pilot may write:
- Allowed actions:
- Prohibited actions:
- Data processing location:
- Retention rule:
- Access owner:

## 5. Evaluation set

- Ordinary success cases:
- Ambiguous-input cases:
- Missing-data cases:
- Permission-denied cases:
- Tool-timeout cases:
- Conflicting-source cases:
- Prompt-injection cases:
- Cases that must go to a human:

For each test case, record the input, expected end state, allowed tools, safety conditions, cost limit and result.

## 6. Economics and limits

- Maximum build budget:
- Maximum monthly operating cost:
- Expected human-review cost:
- Value target:
- Cost per successful task target:

## 7. Delivery and ownership

- Code and configuration owner:
- Required architecture documentation:
- Required test artifacts:
- Monitoring owner:
- Incident owner:
- Handover date:

## 8. Scale-or-stop gate

- Decision date:
- Decision maker:
- Minimum success threshold:
- Maximum acceptable safety failures:
- Maximum operating cost:
- Conditions for expansion:
- Conditions for revision:
- Conditions for stopping:

## Final approval

- Business owner:
- Technical owner:
- Security or privacy reviewer:
- Approved scope date:

Source: Mindela, https://mindela.in/resources/ai-pilot-scope-template