About Architectural Governance
Architectural Governance is an independent advisory practice focused on agentic AI governance, execution-time authority, auditability, and architectural review for high consequence systems.
The work is designed for organisations building or procuring systems where AI may influence or trigger irreversible actions, such as payment release, claims settlement, access control changes, data disclosure, operational dispatch, or other materially binding outcomes.
The focus is not governance as policy language alone. It is governance as executable architecture.
In complex systems, authority is delegated, actions are composed across tools and services, and accountability still remains with humans. Under those conditions, governance either survives at execution time or it collapses, sometimes invisibly, until an incident, audit, insurer review, or regulatory challenge exposes the weakness.
Architectural Governance exists to identify where that collapse can occur, if authority is genuinely enforceable at the point of irreversibility, and if the system can withstand scrutiny.
What this work addresses
Organisations typically engage Architectural Governance when they need defensible answers to questions such as:
- How does authority flow through the system, and can that authority still be evidenced at execution time?
- Where is the real point of irreversibility for each critical action path?
- Can the system refuse inadmissible actions and escalate decisions that exceed delegated authority?
- Will the governance structure withstand audit, incident review, insurer scrutiny, or regulatory challenge?
- When multiple agents, tools, APIs, vendors, or governance domains are involved, where are the structural seams and failure points?
- Are the system's authority model, admissibility rules, evidence requirements, and execution boundaries defined well enough to support safe autonomous or semi-autonomous action?
These are not abstract design questions. They usually surface after deployment, under operational pressure, during procurement, after loss events, or when an external party asks for evidence the system cannot properly produce.
What clients receive
All outputs are written, scoped, and inspection-ready. Typical deliverables include:
- architectural governance and readiness memos
- execution-boundary and authority reviews
- pressure-test and structural risk reports
- authority, admissibility, and evidence gap analysis
- auditability and conformance opinions
- review findings for internal governance, board, insurer, or regulatory use
These artefacts are intended to support:
- engineering and architecture leaders
- risk, compliance, and internal audit functions
- insurers and external auditors
- boards, regulators, and supervisory stakeholders
- organisations assessing whether critical agentic systems are governable before adoption or scale-up
Additional review services
In addition to system level architectural governance work, Architectural Governance provides formal written reviews of existing models, frameworks, papers, and technical proposals.
These engagements are typically requested by:
- founders and product builders developing new agentic systems or governance frameworks
- investors performing technical and governance due diligence on AI companies
- authors preparing papers, whitepapers, or books for publication
- organisations evaluating internal or vendor-proposed architectures before adoption
Typical review scopes include:
- architectural soundness and internal consistency
- authority modelling and admissibility structure
- execution-boundary definition and irreversibility handling
- auditability, traceability, and evidentiary strength
- failure modes, edge cases, and governance gaps
- alignment between claimed guarantees and what can actually be enforced
Outputs are formal written assessments, not informal commentary.
Reviews identify structural weaknesses, over-claims, missing constraints, and areas where governance may fail under real world conditions. Where appropriate, they also outline what would be required to bring a model or system to a defensible standard.
These engagements are paid and scoped, and are typically most valuable once a model, paper, or system design has reached a sufficiently mature stage.
How the work is framed
The practice draws on an architectural governance model built around ISDAIRE, ARETABA, GAG, MGAG, and OTANIS.
These models are used to examine whether:
- autonomous execution is admissible in the first place
- authority remains valid at runtime
- governance survives cross-system composition
- multiple governance layers remain consistent at the point where actions become irreversible
- evidence is strong enough to support audit, review, and accountability
This is advisory work, not operational outsourcing. The objective is to provide independent, written, defensible assessment, not to absorb operational responsibility.
Approach
All work is advisory, human in the loop, and writing-led by default.
I do not implement production systems, deliver code, or assume operational responsibility for live execution. The models and methods used are implementation capable, but the practice itself focuses on independent review, stress testing, governance assessment, and architectural opinion.
Engagements are asynchronous and scoped in writing. This improves precision, creates durable records, and is better suited to complex, high consequence architectural questions than informal calls or quick reactions.
Reviews are paid and formally scoped. There are no free quick looks. That boundary exists to ensure each engagement receives appropriate attention and produces work that is useful, rigorous, and defensible under scrutiny.
About the founder: Dr Masayuki Otani
Dr Masayuki Otani is the founder of this practice. Biography, academic background, OTANIS, professional work, and engagement model are described on the dedicated page.
Ready to discuss?
Contact me with details about your system and the governance questions you face.
Get in Touch