Signed governance receipts
Hosted allow decisions include signed governance receipts that downstream services can verify, and the SDK includes an optional enforcement wrapper for protected handlers.
AI Decision Governance API
Govern AI Decisions Before They Execute
Prime Form Calculus is a governance API that evaluates AI actions against policy rules before execution, verifies authorization, and turns each governed decision into a signed, tamper-evident, and verifiable record.
AI System
↓
Prime Form Calculus
Governance Engine
Governance Engine
↓
Execution Allowed / Blocked
↓
Signed Governance Receipt
↓
Tamper-Evident Decision Chain
↓
Verifiable Audit Record
Prime Form Calculus turns AI decisions into signed, tamper-evident, and verifiable governance records.
Tamper-evident decision chains
Receipts link together in append-only order so teams can detect broken continuity instead of trusting ad hoc logs.
Version-locked decision replay
Stored runtime provenance makes historical decisions replayable against the evaluator version that originally produced them.
Signed and verifiable evidence
Teams can retrieve, verify, and review governed decisions as durable audit records rather than transient application output.
Built for secure AI governance
- Signed governance receipts
- Version-locked decision replay
- Multi-tenant isolation
- API-first architecture
PFC Integrations
Installable Python packages for execution-boundary integrations
PFC is the execution-boundary control plane. The ecosystem access layer now includes installable Python packages for teams looking for a PFC Python package, an execution-boundary Python package, or a practical LangChain governance adapter without replacing the hosted runtime.
Explore PFC integrations for developers to install pfc-core and pfc-langchain, copy working snippets, and understand how framework proposals are intercepted before execution.
From Proposal to Proof
From proposal to proof
PFC governs actions at the execution boundary and proves exactly how an allowed action came to exist.
You can see how decisions become verifiable in the Creative Lineage Demo.
Security at the execution boundary → Learn more
See hosted proof for measured live boundary results across governed load, mixed allow and deny traffic, stale-timestamp freshness enforcement, and recovery after pressure.
AI audit trail explained shows how traceability and execution evidence fit together.
Proposal → Governance → Execution → Proof
Security at Execution
PFC enforces authorization at the moment of action, not just access. This AI security model applies execution-time authorization and runtime policy enforcement as a zero trust execution layer that helps prevent unauthorized actions.
Enterprise-Grade Governance
Prime Form Calculus is designed to help organizations govern automated decisions in environments where policy enforcement, auditability, and operational control matter. Teams mapping the problem space can move from AI execution risk into execution control at the execution boundary.
The first step is recognizing that AI execution risk appears when outputs become actions in production.
Policy Enforcement
Every AI decision request is evaluated against defined governance policies before execution.
Signed and Verifiable Evidence
Each decision remains available as signed evidence in a tamper-evident decision chain, and allow decisions include a verification-ready governance receipt.
Fail-Closed Safety
If a request violates policy or authorization rules, the action is blocked before it can execute.
Tenant-Isolated Infrastructure
Each organization's policies, API keys, and governance records are securely isolated.
Creative Lineage
See how PFC governs creativity
PFC can explore multiple possible actions without granting any of them authority. It evaluates each candidate under live policy, authority, and state conditions, and only an explicitly bound admissible action receives an execution receipt. The system can also prove exactly which signed exploration artifact that execution came from.
Most systems try to filter outputs. PFC proves how an explored possibility became an allowed action.
Output filtering
- tries to control what the model says
- usually acts before or around generation
- often lacks execution-boundary proof
- does not prove how a possibility became an action
Governed creativity with proof
- allows exploration without granting authority
- evaluates candidates under live execution conditions
- binds only explicitly allowed actions
- proves the lineage from exploration artifact to execution receipt
Start Governing AI Decisions
Start governing AI decisions.
Built for teams that need a clear control layer before AI systems act in production environments.