Services
Most engagements start with a paid entry review. Larger reviews are typically commissioned once initial risks and scope are confirmed.
All prices are starting from amounts.
Final pricing depends on system boundary, complexity, and evidence availability.
All work is delivered asynchronously and in writing.
AI Architecture Review and Pressure Testing
We review AI system architecture at the model and system level to determine whether authority, admissibility, and evidence remain coherent under composition. The emphasis is on execution-time governance, non-bypassability of controls, and whether the system's commitment surface is actually controlled. This is an architectural and specification review, not a source code audit.
Pre-Funded Startup Architectural Defensibility Assessment
Duration 3 to 5 working days
Depth Entry-level paid triage
What it is
A focused early-stage review for pre-funded startups building AI systems that may later need stronger governance, clearer authority definition, or investor-facing architectural credibility.
Who it is for
Pre-funded startups preparing for investment conversations, technical due diligence, or early product design validation.
Services included
- High-level system architecture and boundary review
- Identification of obvious governance, authority, and execution-risk gaps
- Early view on whether the system is structurally ready for deeper review
- Recommendation on the next review or design step
Outputs
- Short architectural defensibility memo
- Major structural risks and missing controls
- Recommendation on whether deeper architectural governance work is warranted
Architectural Governance Review
Duration 2 to 4 weeks
Depth Deep independent review
What it is
An independent architectural review of an AI or agentic system to determine whether authority, admissibility, refusal, escalation, and evidence remain coherent at execution time and under system composition.
Who it is for
Organisations building, buying, or reviewing systems where incorrect actions cannot simply be undone.
Services included
- Review of system topology, authority flow, and execution path
- Identification of real irreversibility boundaries
- Assessment of non-bypassability and fail-closed behaviour
- Review of provenance, traceability, and audit survivability
- Governance pressure testing against realistic failure scenarios
Outputs
- Written review report
- Executive summary
- Technical appendix
- Remediation priorities
OTANIS-based Design for Critical Agentic Systems
OTANIS is an architectural governance approach for controlling authority at the moment actions become irreversible. It treats authority as a first-class object and requires explicit admissibility before execution, deterministic enforcement at execution time, and audit-survivable evidence after execution. We design or refactor system architecture so that governance is executable, not merely policy.
- A target architecture that places governance at the point of irreversibility
- Authority object design: scope, delegation, expiry, and provenance expectations
- Execution-time enforcement structure: refusal, halt, escalation, traceability
- A deployment-ready blueprint aligned to the system's real commitment boundary
OTANIS Base Domain Design
Duration 1 to 2 weeks
Depth Domain-level governance specification
What it is
A design package for one governed domain, defining what the system is allowed to do, where authority comes from, where irreversibility sits, and what must be enforced before action can proceed.
Who it is for
Teams defining one critical domain such as payments, claims, access control, booking, dispatch, disclosure, or regulated operations.
Services included
- Domain action classes and scope definition
- Authority source and escalation structure
- Irreversibility mapping and boundary definition
- Admissibility, refusal, and core evidence expectations
Outputs
- Domain governance specification
- Domain risk and irreversibility register
- Executive brief
- Technical appendix
OTANIS Reversible Workflow Design
Duration 2 to 4 weeks
Depth Workflow governance design
What it is
A governance design for one workflow where actions can still be corrected or rolled back, but where authority, admissibility, and refusal still need to be explicit.
Who it is for
Teams moving from advisory AI into action-taking workflows that remain operationally reversible.
Services included
- Workflow decomposition and control-point mapping
- Authority and admissibility rules by stage
- Rollback-aware refusal and escalation logic
- Evidence expectations and governance notes for implementation
Outputs
- Workflow governance specification
- Control-point design
- Evidence and escalation notes
- Implementation-facing guidance
OTANIS Irreversible Workflow Design
Duration 3 to 6 weeks
Depth Critical workflow governance design
What it is
A governance design for one workflow where the system may cross into an irreversible external effect and authority must therefore be enforced at the real point of commitment.
Who it is for
Teams designing or refactoring workflows involving payments, claims settlement, access changes, releases, dispatch, disclosure, or other high-consequence actions.
Services included
- Workflow decomposition and real commit-point mapping
- Authority and admissibility at the execution boundary
- Non-bypassability and fail-closed requirements
- Refusal, halt, fallback, and escalation logic
- Audit and evidence requirements for post-event defensibility
Outputs
- Irreversible workflow governance specification
- Execution-boundary design pack
- Evidence requirements
- Test scenarios and implementation notes
MGAG Multi-Layer Governance Design
Duration 4 to 8 weeks
Depth Cross-layer governance architecture
What it is
A multi-layer governance architecture for systems where one action must remain valid across several organisational, technical, legal, operational, or regulatory layers at once.
Who it is for
Regulated organisations, enterprise platforms, and high-consequence systems where local correctness is not enough.
Services included
- Identification of governance layers and seams
- Layer-specific authority and admissibility structure
- Cross-layer failure and conflict analysis
- Multi-layer refusal and escalation design
- Dependency and evidence expectations across layers
Outputs
- MGAG architecture pack
- Cross-layer governance model
- Conflict and escalation model
- Implementation-facing recommendations
OTANIS Multi-System Orchestration Design
Duration 6 to 12 weeks
Depth Highest architectural depth
What it is
A full OTANIS-aligned orchestration and governance specification for systems where authority must survive across multiple tools, services, rails, and governed action paths.
Who it is for
Organisations building critical agentic systems, platform providers, and enterprises moving high-consequence workflows into serious production architecture.
Services included
- OTANIS-aligned orchestration architecture
- Action classes, candidate boundaries, and governed commit logic
- Authority object and lifecycle expectations
- Refusal, revocation, escalation, traceability, and evidence model
- Cross-system orchestration design for governed execution
Outputs
- OTANIS governance architecture specification
- Orchestration design pack
- Execution-boundary control model
- Evidence and audit design requirements
- Phased implementation guidance
Review of Governance Literature, Papers, Books, and Reports
We review governance-related literature with an architectural lens: what is claimed, what is falsifiable, what depends on hidden assumptions, and what survives contact with real execution boundaries. This includes papers, books, internal reports, and vendor documentation. The output is a critique you can use for decision-making, publication, or product design.
- A structured critique: claims, assumptions, failure cases, and missing definitions
- Identification of where governance is policy-level versus executable control
- Recommendations to strengthen definitions, boundaries, and falsifiability
- A concise executive summary plus deeper technical notes (where needed)
Governance Paper Review
Duration 3 to 5 working days
Depth Single-paper critique
What it is
A structured review of one academic, technical, or policy paper through an architectural governance and execution-time lens.
Who it is for
Researchers, founders, strategy teams, and organisations deciding whether a paper's claims are sound enough to build upon.
Services included
- Critique of claims and stated assumptions
- Assessment of falsifiability and definitional precision
- Identification of execution-time gaps or unresolved boundary conditions
Outputs
- Review memo
- Annotated notes
- Recommendation on whether the paper's claims are defensible in practice
Governance Book Review
Duration 5 to 10 working days
Depth Extended document critique
What it is
A structured review of a book on AI governance, architectural design, or a related domain, examined for internal consistency, falsifiability, and relevance to real execution-time governance.
Who it is for
Researchers, procurement reviewers, strategy leads, and organisations evaluating whether a book's framework or methodology is credible and applicable.
Services included
- Chapter-level or thematic critique of claims and assumptions
- Assessment of whether governance concepts survive execution-time scrutiny
- Identification of structural weaknesses, missing definitions, or policy-only framing
Outputs
- Review memo
- Annotated notes
- Practical assessment of the book's usefulness for design or procurement decisions
Architecture and Governance Red Team Review
Duration 3 to 5 working days
Depth Targeted document critique
What it is
An adversarial review of internal governance, architecture, or strategy documents through an execution-time governance lens.
Who it is for
Enterprise strategy teams, founders, investors, and internal governance groups.
Services included
- Critique of claims and hidden assumptions
- Identification of policy-level versus executable control
- Review of authority model, falsifiability, and execution-boundary logic
Outputs
- Executive critique
- Annotated review notes
- Recommendation on whether deeper work is warranted
Whitepaper Assurance and Technical Credibility Review
Duration 5 to 10 working days
Depth Public-facing technical review
What it is
A structured review of public technical whitepapers, governance frameworks, or architecture claims before publication or commercial use.
Who it is for
Deep-tech startups, platform vendors, founders, and organisations preparing public technical materials.
Services included
- Critique of technical and governance claims
- Identification of hidden assumptions and structural weaknesses
- Review of whether claims survive execution-time scrutiny
Outputs
- Assurance memo
- Annotated review notes
- Optional permissioned attribution subject to approval
How engagements usually start
Most engagements begin with a paid entry review, then move into deeper review or design work once scope and consequence are clear. This keeps work bounded, useful, and proportionate to the system's actual risk.
What this work is not
- No free quick looks
- No source-code audit
- No software implementation or delivery
- No certification of safety or compliance
- No transfer of operational responsibility
How to start
Contact us with details about your system, the governance questions you face, and the service you believe fits your needs. We will respond with a scoping proposal or clarifying questions.
Get in Touch