A new control architecture for consequential autonomous action.
AIgenTXS is designed to bind human and organizational authority to each consequential action, verify that authority immediately before execution, independently reconcile the real-world effect, and preserve proof of the complete lifecycle.
- 01Mandate
- 02Mission
- 03Evidence
- 04ActionIntent
- 05Deterministic decision
- 06Approval or bounded authority
- 07Exact-operation grant
- 08Receiver verification
- 09Observed effect
- 10Reconciliation
- 11Authority response
- 12Audit-completeness proof
The moat is not one dashboard or one model. It is the complete chain and the technical boundaries between its parts.
Most systems address one part of autonomous action.
Hover a category. Each one lights the single link it governs. AIgenTXS connects all twelve into one chain — and the boundaries between them are where the difficulty lives.
Each category is real and necessary. None of them, alone, connects a mandate to a verified real-world effect.
- The mandate model
- Deterministic verification
- Evidence commitments
- Execution-grant infrastructure
- Receiver-side enforcement
- Effect reconciliation
- Audit-completeness proof
- Domain adapters
- Implementation evidence
- Accumulated outcome history
Not one policy engine — the whole set, plus the implementation evidence and accumulated outcome history behind it.
Eight areas. Described by what they accomplish, not how.
hover to hold
Fail-closed candidate promotion
Systems for preventing an AI-generated candidate, recommendation, or proposed action from advancing when required evidence, challenge, supervision, or authority conditions are incomplete.
Unsupported AI recommendations cannot silently become authorized actions.
Claims · schemas · hash constructions · state tables · thresholds · unfiled improvements
Patents are one layer of the moat — not the whole moat.
Seven words that mean different things. We use each one exactly.
The AIgenTXS mandate-to-effect chain has been implemented and adversarially tested in-repository using synthetic fixtures. Product runtime deployment, customer use, and independent attestation are separate milestones and are not implied by patent-pending status.
- ReachedFiledMultiple U.S. applications
- ReachedBuiltMandate-to-effect chain, in repository
- ReachedTestedAdversarially, synthetic fixtures
- Not reachedDeployedSeparate milestone
- Not reachedOperatingProduct runtime — separate milestone
- Not reachedCustomer-usedSeparate milestone
- Not reachedExternally attestedNot implied by patent-pending status
Patent-pending governed-action architecture
SigmaX is built on an architecture designed to govern consequential actions from initial proposal through authorization, execution status, observed effect, reconciliation, and audit proof. Multiple United States patent applications have been filed relating to aspects of this architecture, including fail-closed candidate promotion, evidence-governed authorization, independent effect verification, recovery and supervision, and audit-completeness verification.
Patent pending. No patent has been granted, and no statement on this website represents the scope, validity, enforceability, ownership, or eventual disposition of any patent claim.
Patent pending. Multiple United States patent applications have been filed relating to aspects of the AIgenTXS and SigmaX technology. No patent has been granted. Pending claims may change during examination, and this page does not represent the scope, validity, enforceability, ownership, or eventual disposition of any patent application or patent right. Additional innovations may be maintained as confidential know-how or trade secrets.
- Portfolio-level status only.
- Filing numbers, dates and family maps live in a controlled diligence room.
- File first, explain second.
- Business effect is public; enabling mechanism stays private.
- Governed copy.
- Public IP wording is maintained from one source of truth and reviewed on a dated schedule.