AI Opportunity Canvas

This worksheet helps you figure out whether your company has a process that's ready for a first AI pilot.
Fill out all seven sections to get a clear answer: is this process worth automating?

7sections
~45minutes
1
Process
Which process do you want to evaluate?
Describe one specific, repeatable process that ends with a document or a decision.
2
Process Owner
Who owns this process?
Who decides how it works and is accountable for the results? "Everyone and no one" is a valid answer.
3
Volume & Cost
What does this process cost your company?
Rough figures work fine — exact numbers aren't required.
4
Data
Where is the data and documentation?
This tells us whether a pilot is viable and how soon you can realistically begin.
Other:
5
Three Questions Before You Start
Three questions to answer before you start.
If you can't answer any of these, that's your starting point — not a reason to wait.

Who in the company will sign off on decisions or documents that AI suggests?

What happens if AI makes a mistake — who's responsible, and how will we catch it?

Where will the data going into the AI be stored, and who will have access to it?

6
Measurable Goal
What will we check after 4–6 weeks to know if it's working?
Define a concrete, measurable outcome — not just "it'll be faster," but by exactly how much.
Examples
  • Document prep time drops from __ days to __ days.
  • Error rate decreases by __ %.
  • Information lookup time drops from __ to __ minutes.
My measurable goal →
7
Minimum Scope
Where to start — the smallest meaningful pilot.
One process, one team, one document type. Not "roll out AI company-wide" — just the first concrete step you can complete and measure within 4–6 weeks.
Examples
  • 20 historical documents as a test set — we check whether AI correctly extracts key information.
  • One recurring report type — we check whether AI reduces the time spent drafting the introduction and methodology.
  • Documentation for one process — we check whether employees find answers faster than in paper manuals.
My minimum scope →