INFRA AI Gateway by SPNORTH
Model-neutral control point for governed model access and traceable governance.
Problem
Organizations run several AI models in parallel; access, costs, logs and approvals become opaque – a risk to privacy and traceability.
Intended users
CIO/CISO/CAIO, platform and compliance owners in regulated organizations.
Documented capabilities
- Approved model catalog
- Identity and routing
- Prompt/data-loss controls
- Retrieval authorization
- Logging and evaluation
- Cost quotas
- Red-teaming
- Human escalation
Governance and human boundary
Agents propose; named humans remain responsible for architecture, approvals and regulatory interpretation.
Evidence Flow
Synthetic exampleSynthetic example: illustrated demo with sample data; no customer system and no live telemetry.
- Source
Approved source with version and rights (sample dataset)
- Requirement/Control
Derived requirement with control target and owner
- Human Review
Named approver reviews context, conflicts and evidence
- Test/Evidence
Evaluation against ground truth; reproducible result
- Approval
Signed decision with date and evidence index
Limits and assumptions
- No compliance guarantee
- Maturity status must remain visible
- No productive operation within the readiness scope
Related services
FAQ
Is this available in production?
Status: proposed. SPNORTH does not guarantee compliance, certification, regulatory approval or error-free AI output.
How does this relate to the services?
Relevant services are linked below; entry happens through the documented offers.