AI automation projects often fail before implementation starts. The workflow may be technically possible, but the business case is fuzzy, the handoff is unclear, or the human approval step is missing.
Here is the quick filter I use before scoping one:
The failure-mode question matters most. A demo that works once is not the same as a workflow that can retry, log errors, and hand off cleanly.
A simple ROI estimate:
monthly value = runs per month * minutes saved per run * loaded hourly cost / 60
Then subtract:
If the payback period is not clear, the first milestone should be a smaller diagnostic build, not a full automation.
I packaged this into a small template kit with a calculator, discovery questions, milestone clauses, and a lead qualification checklist. There is also a no-signup web version of the ROI calculator, plus fixed-scope async reviews for teams that want a written milestone/risk map before implementation:
ROI calculator: https://mindtrovertlabs-sketch.github.io/scopegrade-storefront/roi-calculator.html
Workflow Scoping Review: https://mindtrovertlabs-sketch.github.io/scopegrade-storefront/automation-scoping-review.html
Free preview: https://mindtrovertlabs-sketch.github.io/scopegrade-storefront/preview.html
Full kits: https://mindtrovertlabs-sketch.github.io/scopegrade-storefront/
Templates only, no guaranteed savings or outcomes. The point is to make the scoping conversation more honest before anyone spends engineering time.