Define the intent
Name the exact action, target, and purpose. Make the proposal specific enough to check against a mandate.
AIgenTXS / AI Agent Transactions
An agent can propose.
Authority decides.
AIgenTXS is the architecture for governing consequential agent actions — connecting intent, evidence, permission, and the resulting effect in one inspectable chain.
AI Agent TransactionsNot just a payment. An action with intent, authority, and consequence.
Two brands. A connected architecture.
Capital is one application.
The question is universal.
SigmaX Capital asks what should happen before capital moves. AIgenTXS carries the broader architectural question: was this action permitted, did the evidence support it, and did the observed effect match? Each sector supplies its own rules and responsible decision-makers.
From proposal to reviewable proof
Keep the chain intact.
A simplified view of the architecture. The innovation overview explains the full mandate-to-effect lifecycle and implementation evidence.
Bind the evidence
Keep the sources, timestamps, missing items, and counter-evidence attached to the proposal.
Evaluate the policy
Apply the versioned rule and reveal the boundary. A model’s confidence cannot replace a deterministic check.
Verify the authority
Identify the human or bounded organizational authority. Permission for one operation does not become permission for another.
Reconcile the effect
Compare what was authorized with what was independently observed. A completion message alone is not proof of delivery.
Preserve the record
Keep the intent, decision, authority, and effect connected so the lifecycle can be reconstructed and challenged.
An architecture with explicit limits
Built and tested.
Deployment is a separate claim.
The AIgenTXS mandate-to-effect chain is implemented and adversarially tested in-repository using synthetic fixtures. Product runtime deployment, customer use, and independent attestation remain separate milestones.
The public industry examples are local demonstrations of policy concepts. They do not connect to operational systems or grant authority to act.
For teams building with agents
Begin with a boundary
your team can name.
Choose one consequential workflow. Identify the proposed action, its source evidence, the policy it must satisfy, and the person who owns the decision. Use that scope to evaluate how the architecture fits.
The published SigmaX API and TypeScript SDK serve capital workflows. Cross-sector integrations require their own design and review.
A different industry.
The same need for proof.
Designed for consequential actionsCapital. Credit and insurance. Health and life sciences. Workforce and public services. Cyber and critical operations. AI compute and inference. Robotics and physical operations.
One shared architecture, with separate policy, authority, and deployment requirements in every domain.