Evidence
Frameworks for Moving AI Into Real Work
What is published here
Two frameworks, both used in actual consulting and teaching rather than invented for positioning. Each carries a version number, an original diagram, definitions, explicit boundaries covering when not to use it, its relationship to the other, and a citation format. A framework that cannot state where it stops being useful is marketing rather than a working model.
57-word direct answer
Key takeaways
- Two frameworks, deliberately. The count is a constraint, not a starting point.
- Each one states when not to use it, the section most framework pages omit, and the most useful one.
- Both are versioned, so a citation stays meaningful when the model is revised.
- The diagrams are original and reusable with attribution.
The two frameworks
v1.0 · Implementation
Practical AI Adoption Framework
Five stages with explicit exit criteria, for moving a single workflow from observation to a measured result. Answers the question “what do we do next, and how do we know this stage is finished?”
- Problem: pilots that never become operations
- Five stages, each with an exit criterion
- Original cycle diagram
- Used in consulting engagements
v1.0 · Capability
Shuhari for AI Adoption
The Japanese learning progression (shu, ha, ri) applied to how an organization builds AI capability rather than merely buys AI tools. Answers “why did copying someone else's approach not work for us?”
- Problem: tool purchases mistaken for capability
- Three stages of practice
- Original progression diagram
- Used in teaching and workshops
How the two relate
They operate at different scales and answer different questions, which is why both exist rather than one being folded into the other.
The Practical AI Adoption Framework governs a single workflow. It is procedural: five stages, in order, each with a condition that must hold before the next begins. Use it to run an implementation.
Shuhari governs the organization's capability over time, across many workflows. It is developmental rather than procedural: it describes what changes in how a team works as competence accumulates. Use it to understand why a team is stuck, or why an approach that worked elsewhere did not transfer.
In practice an organization runs the adoption framework repeatedly while moving through the shuhari stages. A team in shu should follow the five stages exactly as written. A team in ri will legitimately compress or reorder them, because they understand what each stage was protecting against.
Distinction
The adoption framework tells you what to do next with one workflow. Shuhari tells you whether your organization is yet in a position to depart from it.
What every framework here must carry
This is the publication standard applied to both, and to any added later. It is also a reasonable checklist to apply to anyone else's framework.
A specific problem it solves
Named concretely. “Helps organizations succeed with AI” is not a problem statement.
A version number
So a citation remains meaningful after a revision, and changes are visible.
An original diagram
Hand-authored, information-bearing, reusable with attribution. Not decoration.
Definitions
Every term the model relies on, defined where it is first used.
Steps or components
With the ordering constraint stated, where order matters.
When to use it
The conditions under which it applies.
When NOT to use it
Stated plainly. A model that fits every situation is not a model.
Evidence or examples
Where it has actually been applied, with the limits of that evidence stated.
Relationship to other frameworks
Including honest acknowledgement where it overlaps with existing published work.
A downloadable or printable form
So it can be used away from the site.
A citation format
So attribution is easy and consistent.
Why only two
Naming a framework is cheap, and a long list of proprietary-sounding models is usually a marketing signal rather than an intellectual one. The specification governing this site is explicit: publish only frameworks genuinely used in teaching, consulting, research, or working systems.
Two meet that bar. If a third emerges from real work it will be added; if one stops being used it will be marked superseded rather than quietly maintained.
On originality
Neither framework claims to be unprecedented, and it is worth being direct about that.
Shuhari is a centuries-old concept from Japanese martial arts and traditional arts pedagogy. The contribution here is the application to organizational AI capability, not the progression itself, and the page says so.
The five-stage adoption model overlaps with a good deal of published implementation and change-management practice. Its useful properties are the explicit exit criteria and the insistence on a baseline captured before deployment, not novelty.
A framework does not need to be new to be useful. It needs to be specific, honest about its boundaries, and actually applied.
Frequently asked questions
Can I use these frameworks in my own work?
Yes. Both are published to be used, and each page carries a citation format. The diagrams may be reused with attribution to martinzialcita.com.
If you adapt one substantially, attributing the original and describing your change is more useful to your readers than either silence or strict fidelity.
Is there a downloadable version?
Each framework page is designed to print cleanly to PDF; the navigation, calls to action, and interface elements drop away, leaving the model, diagram, and citation format.
A separately designed PDF with worksheets is planned. It is not published yet, and is listed as outstanding rather than linked.
Which one should I start with?
If you are about to run an implementation, start with the Practical AI Adoption Framework; it is procedural and immediately actionable.
If you are trying to understand why previous attempts stalled, or why an approach that worked at another organization did not transfer, start with Shuhari.