davidfacer.com / aimaturitymodels.com / Strata
Strata
S0–S7 Governed-World Model
Strata defines what makes something real in a governed practice. A document, decision, standard, authorization, system, or act doesn’t become architecture merely because someone created it or stored it in the right place.
A governed thing must answer four questions:
- What kind of thing is it?
- Where does that kind of thing belong?
- What Intent does it realize?
- What authority allows it to act?
A thing with no legal type is undefined. A thing at an illegal stratum is misplaced. A thing with no resolved Intent is ill-typed. A thing without required authority is unactionable.
Why this matters
AI makes explicit a problem architecture has always had. Humans tolerate ambiguity for a surprisingly long time — we infer what a document probably means, remember who approved something, know which copy is current, and recognize from context whether a policy is actually in force. Agents can’t safely operate on those assumptions.
So a governed AI-enabled practice needs more than documents and repositories. It needs a world where the kinds of things that may exist, where they belong, the Intent they realize, and the authority under which they may act are explicit enough to inspect and check. Strata supplies that structure.
It is not a workflow, and S0–S7 is not a sequence of steps. The strata identify different kinds of architectural reality. Click any layer for what it actually governs.
The line running through the layers is the intent fiber — it doesn’t stop at any one stratum, and nothing built here is real without it.
Intent is the fiber
The fiber is not a workflow arrow from S0 down to S7. It is a structural condition present in every valid governed thing. A Standard at S2 must realize Intent. A vocabulary definition at S3 must realize Intent. A work package or authorization must realize Intent. A running system at S6 must realize Intent. An observation at S7 must realize Intent.
Every governed thing is indexed twice over: by where it belongs, and by what it exists to realize. A resolved Intent reference proves that an Intent exists — whether the thing genuinely realizes that Intent remains a matter of human judgment.
Feedback is not the fiber
These are different structures. Fiber is structural: Intent is present throughout every valid stratum. Feedback is dynamic: Observatio may return evidence to Intent so that Intent itself can be reconsidered. A system can be correctly typed and connected to Intent without a feedback event having occurred.
And when feedback does occur, it doesn’t bypass the governed world. An observation that challenges Intent still needs its own valid type, placement, Intent reference, and authority path. Feedback doesn’t mean “change whatever layer is convenient” — it means reality has produced evidence strong enough to reopen a governed question.
Authority crosses the strata
Auctoritas is a stratum, but authority is not confined to S5. S5 holds the formal machinery through which authority is established, delegated, constrained, or issued; the resulting authority governs consequential acts elsewhere. An S6 deployment needs authority. An S2 adoption needs authority. A change to S1 principles needs authority. Retiring a governed artifact needs authority.
So for consequential work the question is never only is this thing structurally valid? It is also is anyone authorized to make it real?
An undertaking is consequential when treating it as real would create, change, approve, commit, deploy, publish, allocate, bind, or retire a governed thing — or otherwise exercise authority. Analysis, explanation, and hypothetical drafting are not consequential merely because their subject is important. Thinking about a contract is not issuing one. Describing a deployment is not deploying. Writing “approved” into a document is not approval.
Structural validity and semantic validity
Strata distinguishes two kinds of correctness. Structural validity asks whether the thing can be resolved: is its type legal, is its stratum legal for that type, does its Intent reference resolve, is the necessary authority present? Those can increasingly be checked mechanically.
Semantic validity asks whether the thing actually realizes the Intent it claims to serve. That remains a judgment. A perfectly typed initiative can still be strategically foolish; a valid Standard can still be the wrong Standard; a properly issued authorization can still authorize a bad decision. Strata makes the structure inspectable. It does not automate judgment out of architecture.
A worked example
Consider an adopted maturity model. The model itself is a Standard at S2 / Normae. Its canonical representation may be stored in S4 / Corpus — but storing it there doesn’t put it in force. Its adoption occurs through S5 / Auctoritas. Its actual use in assessment happens at S6 / Operationes. The findings that use produces become S7 / Observatio, and those findings may improve how the practice operates, change how the Standard is used, cause the Standard itself to be reconsidered — or, in a sufficiently important case, reveal that the underlying Intent needs revisiting.
Throughout that movement the artifacts don’t become interchangeable. A Standard is still a Standard. An authorization is still an authorization. An observation is still an observation. Each remains valid only insofar as it can resolve what it exists to realize.
Better models help. Better governed worlds compound.
Why the models in this family connect
Market intelligence, buyer understanding, and competitive positioning aren’t separate concerns invented independently by engineering and product teams — they’re one shared layer at Normae, feeding into both delivery (the SDLC model) and product decisions (the PDLC model) because that’s structurally where they belong. Feedback velocity shows up in more than one model for the same reason: every model in this family eventually has to answer how fast what it learns makes it back to Intent.