Can this system justify what it is asserting — and can that be independently proven?
SPP2 is the evidence layer beneath your AI systems: every assertion carries a separately attributable verdict, provenance trail, and reproducible receipt. Governance never absorbs a semantic finding, and a semantic finding never absorbs a governance concern.
One evidence engine. A complete platform around it.
Core
The validated semantic evaluation engine — evidence, boundary, identity, and provenance resolved independently of any commercial or governance layer.
Gateway
A versioned commercial API around Core — keys, quotas, rate limits, and signed webhooks — never routed through the engine itself.
Verify
Submit evidence and a proposed assertion. Get back an independent verdict, provenance, and a reproducible receipt.
Monitor
Track a pipeline, agent, or model endpoint over time. Observations are immutable — nothing is silently rewritten.
Enterprise
Isolated deployment boundary, OIDC single sign-on, disaster -recovery-drilled backup/restore, and full audit export.
SDK
Dependency-free Python and JavaScript/TypeScript clients over the same versioned API — nothing proprietary embedded.
Evidence → Meaning → Stability → Authorization → Execution → Proof
Each stage stays independently attributable. A governance concern is never quietly converted into a semantic failure, and a semantic verdict never quietly absorbs a governance concern.