The canonical body of knowledge
Knowledge Base
By Curtis Hibbs and Joshua Barnes
Start with the Applied Enterprise AI overview. It introduces the approach and the questions that guide its development.
This body of knowledge grows alongside the articles, connecting the overview with practical guides and worked cases as they are developed.
Diagnostic Frame
The two tests of effectiveness examine the acceptability of the work and its contribution to the customer outcome, distinguishing that contribution from usefulness to an immediate recipient.
Intervention Engine
Begin with a bounded workflow to make the next action manageable.
Reviewing AI Rules offers a practical way to examine one rule, compare an alternative and follow its effects on recipients and customer outcomes.
Governing Shadow AI: Make hidden AI use governable by connecting candid reporting to authorization, workable alternatives and evidence about the results.
Where Freed Capacity Goes: Decide what the capacity freed by AI should buy, before a budget rule, a staffing tool or the next task decides for you.
Faster Delivery: Find out why a faster step leaves the customer waiting just as long, and what has to change for freed capacity to shorten the wait.
Operating Concepts
Durable design anchors work in explicit outcomes, necessary constraints, and evidence while allowing execution to adapt.
Versioned outcome criteria are revised deliberately and recorded as versions, and they anchor durable design. People refine their criteria as they see outputs, so each design is judged against a named version. Then a change in results can be told apart from a change in criteria. Where it comes from
Customer means the person or organization whose outcome the end-to-end value flow serves. Recipient means whoever receives or uses the output of a particular work step or workflow. A recipient may be an internal colleague, another team, or the customer.
Name the immediate recipient and the customer whose outcome the broader value flow serves. Evaluate the local result, and examine how it contributes to that customer outcome. A useful internal outcome, such as helping a reviewer make a sound decision, matters without by itself demonstrating end-to-end improvement. Make the connection explicit, test it where possible, and identify uncertainty. This can also reveal internal work that is no longer needed.
Two questions sort each AI-enabled change into one of three kinds: did the outcome change, and if not, did the work system change substantially? Acceleration delivers the same outcome through the same work system, only faster, cheaper, or more accurately. Reconfiguration delivers the same outcome through a substantially changed work system. New capability delivers an outcome that was not feasible before. Where it comes from
A work system is the people, AI, and conventional automation that use resources to produce products or services for customers or other recipients. It is the unit of design, bounded by the result it produces. Changes are made at six loci: workflow, task allocation, decision structure, information and feedback, incentives and accountability, and skills. Where it comes from
A minimum viable redesign is the smallest set of changes across the work system that can produce and test a new capability. The minimum is set by the complements: the changes the capability needs in place together. So the capability is named before the redesign is sized. Where it comes from
Vision, path, and means connect the outcome you want, the changes worth testing, and the people, AI, information, and controls that make the work possible.