Three-Layer Provenance
GNS-AIP models the AI supply chain as three distinct layers.
The Three Layers
| Layer | Role | Example | Artifact |
|---|---|---|---|
| Creator | Built the model | Anthropic, OpenAI, Meta | Model origin hash |
| Deployer | Provisions and configures agents | Your company | AgentManifest |
| Principal | Authorizes specific actions | Human user | DelegationCert |
Why This Matters
The Creator has no regulatory obligation to give its models identity. The Deployer, by contrast, has direct regulatory exposure: EU AI Act compliance deadlines, GDPR accountability, FINMA audit requirements.
Don't sell to Layer 1. Sell to Layer 2. Layer 1 has no pain. Layer 2 has compliance deadlines, audit requirements, and liability exposure. Pain is where budgets open.
The Path to Layer 1
When enough deployers use GNS-AIP, the AI labs will integrate at the model level because their enterprise customers demanded it. The path to Layer 1 runs through Layer 2, not around it.