Project 2026

AI in Government Suitability Canvas

Asks whether a public-sector problem should be solved with AI at all, before anyone starts building.

What it is

A structured diagnostic that walks a team through the problem definition, the non-AI alternatives, the data actually available, and the feasibility and risk of the AI option. It is built to produce a defensible no as readily as a yes, which is what separates it from most AI canvases.

Reach for it when

• A team has been told to "add AI" without a problem statement
• A public body needs a documented rationale before committing budget
• An AI proposal needs comparing honestly against a simpler alternative

Running it

1. Define the problem without naming a technology. If it cannot be stated that way, stop here.
2. Write the best non-AI solution properly, not as a straw man. Many sessions end at this step, correctly.
3. Audit the data you actually hold today, not the data you could collect. Be specific about quality, coverage and permission.
4. Close on a written recommendation with its reasoning. A no with reasons is a successful outcome.