AGEPI Nr. 7672 · Designed for OCC SR 11-7 · Basel III IRB · v1.0.0-rs · Controlled Evaluation

Constitutional control for autonomous AI.

SentientROUTER is the governance control plane for AI systems that act on their own. It evaluates every request against a compiled constitution before the model queries any external source, suppresses autonomous external access for high-risk requests, and maintains a dynamic trust budget that automatically de-escalates autonomy when the router itself starts to drift.

See how it works
4/4
Multi-source consensus
3.5 MB
Rust binary
45/45
Automated tests passing
EU
Data residency
7672
AGEPI Patent
Why this exists

Not a router. A constitutional gate.

Most "AI routing" products decide which model to call. SentientROUTER decides whether the system is even allowed to call a model autonomously at all. That is a different problem, and it is the problem that matters in regulated environments where a wrong autonomous decision triggers legal liability.

Problem

Autonomous AI in regulated environments

Banks, insurers, and healthcare systems increasingly deploy AI that takes action — sends emails, issues credits, escalates incidents. The moment action becomes autonomous, the liability question changes.

Gap

Nothing gates before the model is called

Existing guardrails evaluate the AI output. By then, external data sources have been queried, external APIs have been called, and autonomous action may already be in flight. Too late for a regulator.

Solution

Pre-action constitutional evaluation

SentientROUTER classifies every request into S (safe auto), D (draft only), or H (human required) before any external source is queried. H-class requests never reach external models autonomously.

The control plane

How a request flows through SentientROUTER.

Every request passes through a mandatory sequence. This is not a suggestion — it is a compiled-in control plan. If any stage fails its check, the request does not proceed.

Stage 1 · Pre-action

Constitutional evaluation

The request is classified against a machine-readable constitution compiled statically into the executable at build time. This constitution cannot be modified at runtime — only through a documented redeployment. The classification produces one of three classes: S safe autonomous action, D draft for human review, or H human approval required.

Stage 2 · Gate

External source access control

For Class H requests, the router does not query external AI sources in autonomous mode. This is the central technical feature. Even if the downstream model is unrestricted, SentientROUTER prevents the request from reaching it without explicit human approval. For S and D classes, the multi-source aggregation engine activates.

Stage 3 · Consensus

Multi-source aggregation (S/D only)

Three or more independent decision sources (different providers, different models) are queried simultaneously. Each output produces a cryptographic hash included as a separate Merkle node. The final decision is computed via ArgMax with per-source weighting. Divergence is measured via entropy; if it exceeds threshold, the class automatically escalates to H.

Stage 4 · Drift

Router-specific governance drift detection

Unlike generic model drift, this detector measures the router's own behaviour: frequency of S/D/H escalations, divergence trends between sources, distribution of classification outcomes, variation in the dynamic control state over time. Drift detection automatically raises the governance class for affected reflexes.

Stage 5 · Budget

Trust Budget control loop

A dynamic control state variable decreases on high divergence, quarantine debt, policy relaxation without approval, or determinism violation. The feedback loop is explicit: divergence↑ → state↓ → governance↑ → autonomy↓. When the budget reaches zero, all reflexes force-escalate to Class H until human approval restores the budget.

Stage 6 · Execute

Sequential control of execution

Only after all prior stages complete does execution permission resolve. The sequence is mandatory: evaluate → determine aggregation need → establish final class → permit, delay, draft, or block. No parallelism. No shortcuts. Reproducible deterministically via state transition function S(t+1) = F(input, constitution, source outputs, control state).

Stage 7 · Continuity

Offline operation

When connectivity is lost, a persistent local queue signed locally accepts all decisions. On reconnection, atomic synchronization with integrity verification occurs. No decision is lost. No audit gap opens. Governance continuity is maintained independent of network state.

Technical foundation

Built on primitives that earned institutional trust.

SentientROUTER does not invent new cryptography or new policy languages. Like ProvableCORE, it assembles established, battle-tested components into a control plane that banks and regulators can verify themselves.

Runtime

Rust 1.75+ · rustls

Memory-safe systems language. rustls for TLS (no OpenSSL dependency). 3.5 MB binary. Fast cold start on Cloud Run.

Constitution

Compiled statically

Machine-readable rule set embedded in executable at build time. Runtime integrity check via SHA-256 hash verification. Cannot be modified without redeployment.

Consensus

Multi-provider integration

Supports Anthropic Claude, xAI Grok, Google Gemini, OpenAI GPT-4, Groq Llama3, DeepSeek, Mistral, and local Ollama. Each vote hashed independently into Merkle structure.

Divergence

Entropy-based escalation

Divergence between source outputs measured via Shannon entropy or vector cosine. Auto-escalation to Class H when threshold exceeded. Isolated source anomaly reduces its weight and logs as security event.

Audit (optional)

ProvableCORE-compatible

SentientROUTER emits audit events; any compatible cryptographic audit layer can consume them. ProvableCORE is the reference implementation, but the control plane is independent.

Offline

Local queue + atomic sync

Persistent queue with local signing. Atomic reconciliation on reconnection. No decision lost, no gap in audit trail, no weakened governance during network partition.

Two patents, two planes, one architecture

SentientROUTER + ProvableCORE.

SentientROUTER is the control plane: it decides what the AI is allowed to do. ProvableCORE is the proof plane: it creates the evidence that whatever happened is verifiable. They are designed to work together but are independently deployable. Buy one. Buy both. Buy neither and integrate a third-party audit layer that speaks the same protocol.

SentientROUTER

Control plane

Pre-query constitutional gating. H-class external source suppression. Router-specific drift. Sequential control of execution. Offline continuity.

AGEPI Nr. 7672 · 17 April 2026

ProvableCORE®

Proof plane

Tamper-evident Merkle-anchored audit. Decision Proof Objects. Selective disclosure. WORM retention. Third-party offline verification.

provablecore.com → · AGEPI Nr. 7671

The separation is deliberate. SentientROUTER makes sure your AI never autonomously does something it shouldn't. ProvableCORE makes sure that whatever did happen — autonomously or not — can be proven to a regulator, an auditor, or a court. Either on its own solves half the problem. Together they solve the whole problem.

Regulatory alignment

Designed for OCC SR 11-7 and Basel III.

OCC SR 11-7 requires independent model review, effective challenge, and human oversight of autonomous model actions. SentientROUTER is the technical mechanism that makes these requirements real — a compiled rule set that physically prevents autonomous action when the rules demand human intervention.

OCC SR 11-7

Model risk management

H-class routing implements the independent-review and effective-challenge requirements as an infrastructure property. Every autonomous model action that exceeds the approved risk envelope is forced to human review automatically.

Basel III · IRB

Internal Ratings-Based approach

Deterministic state transition satisfies the reproducibility requirements for IRB model validation. PD, LGD, EAD outputs are all governed through the same constitutional gate.

SEC Rule 17a-4

Recordkeeping

Offline queue + atomic sync guarantees that no governance decision is lost during connectivity events. 7-year WORM retention satisfies the SEC recordkeeping requirements when paired with ProvableCORE.

NIST AI RMF

AI Risk Management Framework

GOVERN, MAP, MEASURE, MANAGE functions all have direct mappings to router components. Router-specific drift detection implements continuous MEASURE.

What it looks like

A request, a classification, a decision.

Example response from a SentientROUTER evaluation instance. Format mirrors what a controlled-evaluation deployment returns.

POST /ai/query { "request": "Approve credit line for Farm ID 73410, EUR 250,000", "context": "high-risk-credit" } → 200 OK { "class": "H", "reason": "Credit approval > EUR 100k requires human review per constitution v2.0 rule #4", "autonomous_action_permitted": false, "external_sources_queried": false, "jury": "not_invoked", "escalation_required": true, "approval_endpoint": "/governance/approve/req_8f3a2...", "audit_event_id": "evt_7b1c9...", "constitution_hash": "e691cd98..." }

Notice what did not happen:

Common questions

What banks, regulators, and boards ask first.

We already have guardrails on our AI. Why do we need a separate control plane?
Guardrails evaluate outputs. They happen after the model has already been called, after external sources have already been queried, after autonomous action may already be in motion. SentientROUTER operates before the model is called. For Class H requests, the external AI provider never even sees the request. That is a different kind of guarantee, and it is the guarantee regulators look for under OCC SR 11-7 and EU AI Act Article 14.
Does this slow down our AI pipeline?
Constitutional evaluation is fast — milliseconds, compiled rule set, no network calls. The slowdown happens when we block something that should be blocked. That is the point. If your operational concern is latency for Class H decisions, the correct response is not to remove the gate but to reduce the set of operations classified as Class H.
What if the constitution itself is wrong?
The constitution is compiled into the executable. Changing it requires a documented redeployment — new build, new version, new deployment artefact. This is deliberate: the constitution is not a runtime setting that can be modified under pressure. If the constitution is wrong, you fix it and redeploy, and every historical decision can be replayed against the new version in the counterfactual simulator to see what would have changed.
Is SentientROUTER proprietary or open?
Proprietary. Patent application filed under AGEPI Nr. 7672; USPTO provisional pending. Commercial licence required for commercial deployment. The audit protocol it emits is open and ProvableCORE-compatible, so you are never locked into a single audit vendor.
Can we use SentientROUTER without ProvableCORE?
Yes. SentientROUTER emits audit events in a format compatible with any tamper-evident audit layer. ProvableCORE is the reference and the easiest integration, but not a requirement. Existing SIEM, audit log, or compliance pipelines can consume the events directly.
What happens when your VPC loses connectivity?
The offline module activates. All decisions are signed locally and queued in a persistent local store. On reconnection, atomic synchronization with integrity verification re-joins the main audit trail. No decision is lost. No governance gap opens. For regulated environments this is non-negotiable — network issues must not become compliance issues.
How does the Trust Budget decide when to restore autonomy?
The budget increases only on verified good performance and decreases on governance anomalies. When depleted, it forces all reflexes to Class H. Restoration happens only through an explicit human approval event with domain `trust_budget_restore` — there is no automatic recovery. This prevents the system from silently returning to autonomous mode after a period of poor behaviour.
Is there a free tier?
A 10-day controlled-evaluation pilot is available for qualified financial institutions. For longer trials or non-commercial evaluation, contact us directly. Commercial deployment requires a commercial agreement.

Evaluate SentientROUTER on your own AI system.

10-day pilot. Full feature access. Integrated with ProvableCORE or your existing audit layer. Deployed in US or EU data residency.