Sovereign Core v1.0.0

Sovereign Core — Intelligence That Belongs to You

A deterministic-first cognitive engine inside AEGIS. Governance, verification and orchestration run without any AI dependency. When AI helps, it is invoked — never required. No GPU needed. No vendor lock-in. No internet required.

AI as escalation, not foundation.

Sovereign Core is the cognitive kernel of AEGIS — the layer that decides how every request is solved. Most platforms answer everything with a large language model. Sovereign Core does the opposite: it solves each request at the lowest, most reliable layer capable of solving it, and escalates upward only when necessary. The result is an autonomous operations platform whose control plane is fully deterministic, fully auditable, and fully operational even with AI switched off entirely.

AI as escalation, not foundation

Deterministic rules, symbolic checks, a structured knowledge base and lightweight statistical models resolve the majority of work. Generative AI is the top tier — optional and bounded.

Fail-safe by default

Out of the box, the system runs in CORE mode: every AI capability refuses. You switch capability on per policy, never off.

Every decision leaves evidence

Each invocation records what ran, where, how fast and why — in a tamper-evident, hash-chained ledger. Raw payloads are never stored.

The Escalation Ladder.

Every request climbs only as high as it must:

LayerWhat it doesNeeds AI?
1. DeterministicPolicy evaluation, access control, schema validation, state machines, scheduling, drift detection, approvalsNo
2. SymbolicRule reasoning, dependency analysis, constraint checks over the knowledge graphNo
3. KnowledgeStructured world-state queries: actors, services, permissions, deployments, tenantsNo
4. StatisticalEmbeddings, classification, similarity, ranking, anomaly detection — small models on CPULightweight local models only
5. GenerativeCode generation, summarization, open-ended reasoning — bounded, budgeted, and its output always re-enters verificationOptional; policy-gated
6. HumanMandatory approval gates and unresolved ambiguity
Key invariant: generative output is never its own validator. Anything the top layer produces must pass the deterministic verification pipeline before it becomes evidence.

One platform, five postures.

Switching is a policy decision, not a redesign.

ModeAINetworkUse case
COREOffOffRegulated, air-gapped, maximum-audit environments
DEGRADEDOffOffIncident or contingency operation
LOCAL AIOn-device onlyRestrictedData-sovereign deployments, CPU-only infrastructure
ACCELERATEDOn-device + acceleratorsRestrictedHigher throughput when hardware exists
HYBRIDOn-device + externalFullMaximum capability under tenant policy
CORE mode is not a reduced product: governance, software factory verification, evidence vaults, approval gates, compliance reporting and orchestration remain fully functional with zero AI and zero connectivity.

Independence you can operate on.

Auditability

Deterministic decisions are reproducible and replayable. Every AI call is budgeted, hash-recorded and inspectable. Non-deterministic components are never placed in trust positions.

Sovereignty

Capability contracts are vendor-neutral: the platform requests “embed” or “classify”, never a specific provider. Providers are configuration, not architecture.

Cost proportionality

Compute matches the actual cognitive requirement. Most workloads resolve on CPU in milliseconds; heavy models are reserved for the few tasks that genuinely need them.

Resilience

Connectivity loss, vendor outage or hardware absence degrade capability — they never take governance down.

Security

Removing language models from the control plane eliminates prompt-injection surface where it matters most: approvals, evidence and enforcement.

Measured, not promised. (v1.0.0)

Every figure below comes from the shipped v1.0.0 runtime and its regression gate.

5 modes

Five-mode runtime (CORE / DEGRADED / LOCAL AI / ACCELERATED / HYBRID), fail-closed default enforced in production.

384-dim · ~80 ms

CPU-first embedding pipeline: 384-dimensional vectors at roughly 80 ms per call on standard server CPUs — no GPU anywhere in the path.

SHA-256 chain

Tamper-evident evidence chain: SHA-256 linked events, verifiable end to end.

Payload-free

Payload-free telemetry: invocation ledgers store metadata and hashes only.

15 / 129 tests

Capability contract suite: 15 dedicated conformance tests inside a 129-test green regression gate.

Multi-tenant

Tenant isolation inherited from AEGIS: tenant scope resolved from signed identity, never from request bodies.

Short answers, straight.

Does Sovereign Core replace AI?+
No. It right-sizes it. AI remains available where it creates real value — and becomes unnecessary everywhere else.
What if we have no GPU at all?+
That is the design baseline. The full governance plane runs on commodity CPUs.
Can we run it fully offline?+
Yes. CORE mode operates with networking disabled while retaining governance, verification, evidence and compliance functionality.
Is generative AI blocked?+
It is gated, not banned. Tenant policy decides when the generative tier may be invoked, within budgets — and its output is always re-verified deterministically.
How do you avoid vendor lock-in?+
The platform speaks capabilities, not vendors. Any conforming provider — local or remote — can serve a capability without touching application code.

Autonomy without dependence. Intelligence with receipts.

Sovereign Core turns architectural independence into an operational guarantee: your governance keeps working when models change, vendors disappear, networks fail or GPUs are simply not there.