Service · Artificial intelligence
AI where it changes the economics
Most AI programmes stall in the same place: a portfolio of pilots nobody can rank. The work is deciding what is worth doing, proving it inside the operating business, and stopping the rest early.
Problems this solves
- A pilot portfolio that needs ranking by value at risk rather than by enthusiasm
- Digital transformation that has to survive contact with an operating business, not a slide
- Product evolution where AI changes the product itself rather than the way it is built
- Process optimisation with the savings stated as a number someone will be held to
- A build-or-buy call on models, tooling and data that will still look sensible in two years
What you can check before you call
The working method for this line is published: a model for ranking an AI portfolio by what it costs to find out, with the blocking rules and the kill gate we use, free and without an email address. Its worked example is built from public data. Behind it is the same method that produced the telecom work, down to what AI does at each step and what a person decides.
Where this usually starts
The rungs
R1
One-day teardown
- A 90-minute working session
- A written answer in two working days
- The model behind it, yours to keep
Indicative scope
R2
Two-week assessment
- Two working sessions
- A 12–18 page memo, working shown
- A one-page board summary
Indicative scope
R3
Project
- Scoped per engagement
- Weekly checkpoints
- All models and sources at close
Indicative scope
The proof is the method, the tools and the analyses, all public and all auditable.