Maturity models do not get funded
Most AI strategies arrive at the board as a five-stage maturity curve with the organization marked somewhere around stage two. It is accurate, it is well-researched, and it does not move money. What moves money is a thing that works, in production, that someone can point at.
So the first job of an AI strategy is not to describe the destination. It is to identify the single workflow that can be shipped inside one budget cycle and will be visible to the person who controls the next one.
Four tests for the quick win
Volume you can count. If nobody can tell you how many times the workflow runs per month, you will not be able to prove the impact either.
A single owner. One person who can say yes to a process change without convening a committee.
A metric that already exists. Do not invent a measurement to prove your project. Use handle time, cost per contact, cycle time, something already on a dashboard someone reviews.
Tolerable failure. Pick a workflow where a wrong answer is an inconvenience, not an incident. Save the clinical and financial contexts for after you have credibility.
What to do with the rest of the roadmap
It still matters, but it belongs behind the quick win, framed as what the win unlocks. Governance structures, funding paths, and readiness benchmarking are far easier conversations when they are the operating model for something already running.
Takeaways
- Pick one workflow with countable volume, a single owner, and an existing metric.
- Frame the wider roadmap as what the first win unlocks, not as the ask itself.
- Tolerable failure beats maximum impact for a first engagement.