How the scorecard works
Each answer scores 0–4. A dimension's score is its answers as a percentage of the maximum, and the overall score is the weighted average. The weights favour value, but a high-value workflow with no owner still won't be recommended.
| Dimension | Weight | Questions |
|---|---|---|
| Value | 35% | How often does the workflow run? · Effort per run, all people combined · How many people do this work? · If it were faster or better, what would change? |
| Feasibility | 25% | Where do the inputs live? · What do the inputs mostly look like? · How much does each run vary? · Can someone check an output quickly? |
| Risk (higher = safer) | 20% | If an output is wrong… · Most sensitive information involved · What should the AI do? |
| Readiness | 20% | Who owns the workflow? · What AI tools does the team already have? · Do you know what the workflow costs today? |
How the approach is chosen
The recommended approach doesn't come from the score alone. The rules below are checked in order, and the first one that matches wins:
- Not yet — fix ownership first — the workflow has no named owner.
- Don't pursue this one now — value scores under 35.
- Not yet — fix the process or data first — the inputs are offline or tacit, or outputs can't be checked.
- Deterministic automation — no generative AI needed — the steps are identical every time and the inputs are structured.
- AI-assisted only with strong controls — review readiness first — risk scores under 40, outputs could cause harm, or the AI would act on its own.
- Configure the tools you already have — you already license an AI assistant and it would only draft or suggest.
- AI-assisted workflow — bounded and human-reviewed — everything else.
This is a screening tool, not a business case. Use it to compare candidates and decide what to investigate. Then work out the numbers with the economics calculator, and write them up with the brief builder. For the full selection method, see Where should our business start with AI?
Disclosure:Paid engagements are contracted and delivered by Neul Labs, which Dipankar Sarkar founded and leads. Playbook recommendations stay vendor-agnostic, and we do not publish supplier rankings.