Sovereignty
Sovereignty in TramAI means that AI execution stays under explicit organizational control. Models, providers, tools, routes, approvals, artifacts, and audit outputs are not left to convention or best effort. They are described, constrained, and enforced as part of the runtime itself.
This matters because production AI systems are no longer just inference clients. They move sensitive data, make routing decisions across trust boundaries, invoke tools, resume workflows, and create outputs that may need to stand up to internal review, customer scrutiny, or regulatory inquiry.
TramAI treats sovereignty as a product capability, not a deployment footnote.
What Sovereignty Means In TramAI
- Explicit control over execution: allowed models, providers, tools, and permissions are declared rather than assumed.
- Trust-boundary awareness: data classification and provider trust zones are part of routing decisions.
- Fail-fast governance: misconfiguration is rejected at startup or build time instead of surfacing as a runtime surprise.
- Auditability by default: policy decisions, approvals, and verification results can be retained as evidence.
- Offline and local-first support: teams can run with local-only providers or fully air-gapped deployment rules when required.
The current release is 0.6.0. Sovereignty in 0.6.0 is configuration, not code: the same unified Spring Boot starter runs either the standard or the sovereign runtime, selected by tramai.profile. The sovereign modules are published preview API — not a frozen 1.0 surface — and TramAI makes no compliance certification claim; it provides enforceable runtime controls and audit-grade evidence that your own review process can assess. Start with the Sovereign Runtime Quickstart.
Why This Is Important
The value of sovereignty is not limited to heavily regulated environments. It also matters when a company:
- needs to know exactly where prompts and context are allowed to go
- cannot allow arbitrary tool execution from model output
- wants repeatable controls across multiple products or tenants
- needs deterministic evidence for procurement, compliance, or customer security reviews
- wants the option to run local models without losing structured runtime guarantees
In This Section
| Page | Focus |
|---|---|
| What Sovereignty Means | The conceptual model and how TramAI defines sovereign AI execution |
| Who Should Use It | Teams, companies, and deployment profiles that benefit most |
| Operational Implications | What adopting sovereignty changes in architecture, process, and runtime operations |
Related Technical Documentation
The sovereignty story is supported by the implementation documented under security:
- Sovereign Runtime Quickstart
- Sovereign Mode
- Approval Workflows
- Artifact Verification
- Evidence Packs
- Offline Deployment
If security explains the mechanisms, this section explains the strategic meaning of those mechanisms for users and organizations.