Home/Frameworks
Frameworks
Working models for deciding what an AI system may do. Published in full, free to use, with the failure modes included.
These are the models I use with architecture and leadership teams, and the spine of the talks. They are here in their entirety rather than as a teaser, because a method you can apply in fifteen minutes tells you more about whether I am useful than anything I could claim about myself.
-
Capability, authority, accountability
Three questions to ask of any AI proposal that reaches your desk. What can it do, what may it do here, and who answers for the outcome. Most proposals answer the first and quietly skip the other two.
Read the framework → -
The delegation ladder
Read, draft, send, commit. Four rungs of autonomy, what each one costs when it is wrong, and how to notice a system standing on a rung it was never assessed for.
Read the framework →
Using these
Take them. Use them in your own decks, your own governance documents, your own arguments. Attribution is welcome and not required. If you find a case where one of them breaks, I would genuinely like to know, and I will say so on the page with the date on it.
How I label the claims on this site, and why there are no client numbers on it, is set out in how I make claims.
If you want these applied rather than adapted
These are the models. Running them against a real architecture, with your constraints and your evidence, is the work I do with teams.