Formal safety models, ODD specifications, assurance cases, the ERI Prediction continuation-in-part brief, and non-technical reviewer briefs. Circulation is private. Request a copy and we will send the file directly.
The Assured Autonomy 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.
System architecture, novel elements, patent claim overview
Executive summary of the Assured Autonomy Governor: architectural overview, five-stage pipeline (Observe → Assess → Modulate / Anticipate → Enforce → Audit), bounded-authority contract, novel elements versus prior art (OPA, MLflow, model cards, OpenTelemetry), and the scope of the patent claims.
Assurance case, formal properties, test evidence
Formal safety model and assurance argument for the Assured Autonomy 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 Assured Autonomy 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 / Anticipate → 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.
Prior-art comparison, novelty argument, safety claim tree
Structured assurance case for the Assured Learning Governor: prior-art landscape (MLflow, model cards, OpenTelemetry, drift-detection systems), novelty claims for divergence-based learning-event authority modulation with rollback contract, and the safety claim tree that ties learning-time governance to the patent’s protected constructs.
The ERI Prediction layer extends the runtime governor with an anticipatory step. Substrate state projects forward at multiple horizons, per-horizon predictions seal alongside the current verdict, and reconciliation receipts at horizon make the predicted-vs-realized comparison cryptographically auditable. Continuation-in-part filing on Patent A is imminent. This page is the public declaration of the work in advance of the filing.
System architecture, novel elements, patent claim overview
Executive summary of the ERI Prediction layer: architectural overview, per-horizon predictions sealed alongside the current verdict, the Modulate / Anticipate insertion point in the runtime pipeline, novel elements versus prior art, and the scope of the continuation-in-part claims. The CIP filing on Patent A is imminent; this one-pager is the public declaration of the work in advance of the filing.
ERI Prediction · Formal model and test evidence
Formal safety model for the ERI Prediction layer. Describes the substrate state observables (Energy, Resonance, Information), the per-horizon predictions sealed inside each governance record, the operator-hash binding, the trajectory-modulation rules that downgrade verdicts before the action commits, and the reconciliation contract between predicted and realized records at horizon. Empirical validation on real cyber benchmark data inside the patent latency baseline.
Predictive ODD · Formal envelope for trajectory-aware authority modulation
Operational Design Domain specification for the predictive layer. Defines the formal envelope within which per-horizon predictions are admissible, the drift / cascade / escalation thresholds that drive trajectory-aware downgrade, the reconciliation contract between predicted and realized records at horizon, and the audit-evidence structure regulators read.
Prior-art comparison, novelty argument, safety claim tree
Structured assurance case for the ERI Prediction layer: prior-art landscape (Koopman operator learning, dynamic mode decomposition, runtime trajectory monitors), novelty claims for cryptographically sealed per-horizon predictions with operator-hash binding and reconciliation receipts at horizon, and the safety claim tree that ties anticipatory governance to the continuation-in-part claims.
Plain-language briefs for CTOs, CISOs, and technical reviewers who need the shape of the two filed patent families and the predictive-layer continuation-in-part 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 filed patent families and the continuation-in-part on the predictive layer do, why they matter now (EU AI Act, NAIC Model Bulletin, FDA PCCP), what is novel versus existing infrastructure (MLflow, OPA, OpenTelemetry, drift-detection systems), and the comparison table. No jargon. Designed to be reviewed over a coffee.
Architecture, decision grammar, hash-chain design, sealed forecasts
The deeper technical follow-up for reviewers who want architecture. Covers decision grammar, governance-step result structure, hash-chain construction, the ERI Prediction layer (per-horizon predictions, operator-hash binding, reconciliation receipts at horizon), the structural-inaccessibility claim, and an extended prior-art comparison.
Reply with your affiliation. We send the file directly.
