PLANCKCYBER — AI WORKFLOW SCREENING CHECKLIST

Use this as a first-pass screening tool. It is directional, not a guarantee that AI is appropriate.

1. PAIN & FREQUENCY
- Does the workflow happen often?
- Does it consume meaningful skilled time or create delays/rework?

2. BOUNDED WORK
- Can inputs, outputs and exceptions be described?
- Is there a clear point where a person should review or take over?

3. DATA & SYSTEM ACCESS
- Is the required information accessible and reasonably reliable?
- Can the relevant systems be reached through suitable APIs or controlled interfaces?

4. MEASURABLE ACCEPTANCE
- Can you define task quality, cycle time, exception rate, cost, adoption or another meaningful measure?

5. RISK & REVERSIBILITY
- Can mistakes be detected and corrected?
- Are permissions and high-impact actions appropriately constrained?

6. OWNERSHIP
- Is someone accountable for the workflow and the decision to implement it?

A strong first opportunity is usually painful, repeated, measurable, technically accessible and safe enough to test with a focused proof.