Formal safety models, ODD specifications, assurance cases, and non-technical reviewer briefs. Circulation is private — request a copy and we will send the file directly.
The Actuation Governor bounds the authority of a running autonomous system. It enforces rate, scope, and safety-envelope limits on the decisions reaching actuation — and signs every one of them to a tamper-evident audit chain. AAG papers are in the process of conversion from the originals; drop a note and we'll notify you on release.
Assurance case, formal properties, test evidence
Formal safety model and assurance argument for the Actuation Governor. Describes proof boundaries, five-stage pipeline, runtime invariants, and cross-platform test results. Intended for certification reviewers and safety engineers.
Envelope definition, determinism guarantees, audit evidence
Formal ODD specification for runtime actuation governance. Defines bounded authority envelope, deterministic behavior under adversarial input, state-machine contracts, and hash-chained audit evidence format.
Prior-art comparison, novelty argument, safety claim tree
Structured assurance case for the Actuation Governor: prior-art landscape (OPA, MLflow, model cards, OpenTelemetry), novelty claims, and the safety claim tree that ties runtime behaviour to the patent’s protected constructs.
The Assured Learning Governor bounds the authority of parameter updates — every learning step is read against an explicit envelope, modulated where necessary, enforced where required, and signed. The result is an adaptive system that regulators can still read.
System architecture, novel elements, patent claim overview
Executive summary of the Assured Learning Governor: architectural overview, five-stage pipeline, mode-management contract, novel elements versus prior art, and the scope of the patent claims.
Learning-time governance — formal model and test evidence
Formal safety model for the Assured Learning Governor. Describes the five-stage Observe → Assess → Modulate → Enforce → Audit pipeline, safety properties LC0–LC7, and 87/87 cross-platform test evidence supporting the certification claim.
Learning ODD — formal envelope for parameter-update authority
Operational Design Domain specification for learning-time governance. Defines the formal envelope within which adaptive model parameter updates are permitted, the deterministic rollback contract, and the audit-evidence structure regulators read.
Plain-language briefs for CTOs, CISOs, and technical reviewers who need the shape of the two patent families without wading into claim language. Written to be read in ten minutes.
Two-page gut-check for non-domain CTOs and CISOs
The plain-language two-page framing: what the two patent families do, why they matter now (EU AI Act, NAIC Model Bulletin, FDA PCCP), what is novel versus existing infrastructure (MLflow, OPA, OpenTelemetry), and the comparison table. No jargon. Designed to be reviewed over a coffee.
Architecture, decision grammar, hash-chain design
The deeper technical follow-up for reviewers who want architecture. Covers decision grammar, governance-step result structure, hash-chain construction, the structural-inaccessibility claim, and an extended prior-art comparison.
Reply with your affiliation. We send the file directly.
