Every AI response is scored sentence by sentence against your sovereign documents — contracts, policies, regulations. If a claim isn't grounded in your knowledge base, the connection is cut. Automatically.
When an AI tells a customer the wrong interest rate, contradicts a contract clause, or invents a policy that doesn't exist, the consequences aren't just reputational. They're measurable.
Each unverified AI response that reaches a banking customer becomes a potential regulatory event with mandatory resolution deadlines.
Traditional hallucination checkers evaluate responses after delivery. The damage is already done. You need interception before the message arrives.
Sending every AI response through a second LLM for verification doubles cost and adds unbearable latency for real-time applications.
The runtime opens a streaming channel to the LLM. Each sentence is accumulated, scored against your knowledge base, and either passed or intercepted — before it exits the system.
A persistent HTTP stream is established with the LLM. Tokens arrive in real time as the model generates them.
Tokens are buffered and assembled into complete semantic units — sentences or logical clauses — before scoring begins.
Each sentence is compared against the chunks most relevant to the query, retrieved from your on-premise knowledge base. A semantic integrity score is computed: how well-supported is this claim by your actual documents?
A running heat metric tracks how much the model has been drifting across recent sentences. A single low-score sentence may pass; a pattern of drift triggers earlier intervention.
If the collapse velocity crosses the threshold — accounting for score, heat, and document entropy — the HTTP connection is terminated. The hallucinating sentence never reaches the user. A citation-grounded substitute can be injected instead.
Every response gets a Trust Score (0–1), source citations with clause references, and an immutable audit log entry. Compliance teams can replay any incident.
You don't have to change your LLM provider. The runtime intercepts at whichever layer gives maximum control for your setup.
Works with any closed-source LLM — cloud APIs, enterprise endpoints — without requiring access to model weights or GPU. Intercepts at the HTTP streaming layer.
When you run local model weights directly, interception goes deeper — into the probability distribution the model produces before each word is chosen. The steering happens before the token is even generated.
Every enterprise has ground truth: loan contracts, insurance policies, operational procedures, safety regulations. Those documents are loaded into the runtime as an immutable reference.
The AI is allowed to answer freely — but every sentence it produces is compared against that reference in real time. Not by another AI. Not by keywords. By semantic similarity to the actual chunks of your actual documents.
Every response gets a numeric Trust Score (0–1) reflecting how well-grounded it is against your documents. Exportable, auditable, traceable.
Each verified sentence references the exact source: document name, section, clause, and page. Your users always know where the answer comes from.
When cumulative drift across sentences crosses the threshold, the HTTP stream is terminated. No hallucinated content exits the system.
Before deploying your AI, the system scans your own documents for internal contradictions. Catch inconsistent clauses before the LLM finds them at your expense.
Prompt injection, PII leakage, jailbreak attempts, and indirect injection via documents are blocked before the AI even sees the request.
Every interception, every Trust Score, every citation is logged with timestamps. Regulators can inspect the full decision history for any session or incident.
Three lines of code to wrap any LLM call. Native callbacks for LangChain, LangGraph, and any OpenAI-compatible endpoint. No infrastructure rewrite needed.
No model calls to external APIs. No data transmitted to third-party services. Your sovereign documents and customer queries never leave your infrastructure.
| Capability | Rules / Regex | LLM-as-Judge | RAG-only | This Runtime |
|---|---|---|---|---|
| Interception point | Post-generation | Post-generation | Pre-generation (retrieval only) | Mid-stream, per sentence |
| Requires your documents | No — generic patterns | No — LLM knowledge | Yes | Yes — clause-level |
| Latency overhead | <5ms | 2,000–8,000ms | ~50ms (retrieval) | ~14ms |
| Understands context drift | No | Partial | No | Yes — thermal tracking |
| Clause-level citations | No | Sometimes | Document-level | Section + page |
| Blocks mid-stream | No | No | No | Yes — circuit breaker |
| On-premise, no cloud | Depends | Requires LLM API call | Depends | Yes — fully isolated |
| Audit trail with clause mapping | No | No | No | Yes — immutable log |
| Pre-deploy contract audit | No | No | No | Contradiction detection |
Loan officers, customer-facing chatbots, and internal compliance assistants all operate under strict regulatory frameworks. A hallucinated rate, a made-up deferral policy, or an incorrect fee structure triggers mandatory complaint resolution and potential sanctions.
The runtime is pre-mapped to CONDUSEF resolution categories, CNBV disclosure requirements, and LTOSF article structures. Every blocked response includes the specific article it would have violated.
Operational AI assistants that misquote safety procedures, maintenance thresholds, or environmental compliance requirements don't just create liability — they create physical risk. The sovereign document here is the operational procedure manual, not a contract.
Works with technical document formats: maintenance manuals, ISO standards, safety regulations. Any document you can load is the baseline the AI must stay faithful to.
Coverage interpretations, exclusion clauses, and premium calculations are the territory where AI most commonly invents things that sound plausible but contradict the actual policy wording. The runtime loads the policy documents as sovereign truth.
Built-in support for multi-document truth sources: main policy, endorsements, exclusion riders. Clause references in citations point to exact policy sections, ready for dispute resolution.
30 pairs of real/hallucinated responses across 6 banking contract categories, tested against the actual scoring engine. The numbers below come from benchmark/results.txt — not a simulation.
Un segundo contrato de crédito personal fue cargado como Prisma Estático en formato de cláusula completa en prosa legal. El governor se calibró sobre 30 respuestas etiquetadas (calibration set) y luego fue evaluado ciegamente sobre 40 respuestas nuevas (hold-out set) — nunca vistas durante la calibración.
Separación estricta calibración/prueba. El threshold óptimo se determina solo sobre los 30 casos de calibración y se aplica fijo sobre los 40 de prueba. Cero data leakage. Estándar universal en Machine Learning.
Pares gemelos donde solo cambia un dato concreto — número, porcentaje, titular, plazo. Detectar diferencia fina (18.5% vs 16.9%) es mucho más difícil que diferencia gruesa (18.5% vs 35%). Referencia: Gardner et al., EMNLP 2020 — Contrast Sets.
Los cuatro números que el mercado reconoce como estándar absoluto: TP · TN · FP · FN. De ahí se derivan Precision, Recall, F1 y Health. Reportamos todos los números, sin ocultar fallos.
| ID | SI | Veredicto Governor | Etiqueta Real | Respuesta |
|---|---|---|---|---|
| C2-HO-001 | 0.1864 | PASS | CORRECTA ✓ | Tu tasa de interés anual es del 18.5% fija… |
| C2-HO-002 | 0.2250 | PASS | CORRECTA ✓ | El CAT de tu crédito es 24.7%… |
| C2-HO-003 | 0.1429 | PASS | ALUCINACIÓN ← FN | La tasa es 18.5%, pero puede subir… |
| C2-HO-004 | 0.0250 | BLOCK | CORRECTA ← FP | No hay penalización por pago anticipado… |
| C2-HO-005 | 0.1111 | BLOCK | CORRECTA ← FP | La comisión por apertura es del 2%, $2,400… |
| C2-HO-006 | 0.1111 | BLOCK | ALUCINACIÓN ✓ | Pago $3,580 ya incluye el seguro… |
| C2-HO-007 | 0.1429 | PASS | CORRECTA ✓ | Si te atrasas, la comisión es $350 por evento… |
| C2-HO-008 | 0.0851 | BLOCK | ALUCINACIÓN ✓ | El seguro de vida es opcional… |
| C2-HO-009 | 0.1404 | PASS | CORRECTA ✓ | Plazo 48 meses, corte día 15… |
| C2-HO-010 | 0.2368 | PASS | CORRECTA ✓ | Tasa moratoria 2.5× ≈ 46.25% anual… |
| C2-HO-011 | 0.0732 | BLOCK | CORRECTA ← FP | El monto de tu crédito es de $120,000 MXN… |
| C2-HO-012 | 0.0682 | BLOCK | ALUCINACIÓN ✓ | Anticipado con comisión del 3%… |
| C2-HO-013 | 0.1500 | PASS | ALUCINACIÓN ← FN | El CAT es del 22.4%… |
| C2-HO-014 | 0.1571 | PASS | CORRECTA ✓ | Tasa 18.5%, con seguro CAT sube a 24.7%… |
| C2-HO-015 | 0.1489 | PASS | ALUCINACIÓN ← FN | $350 más IVA por evento… |
| C2-HO-016 | 0.0833 | BLOCK | CORRECTA ← FP | Seguro vida obligatorio $85 adicionales… |
| C2-HO-017 | 0.0649 | BLOCK | CORRECTA ← FP | $120,000 en 48 mensualidades de $3,580… |
| C2-HO-018 | 0.1277 | BLOCK | CORRECTA ← FP | Día de corte el 15, pagar antes… |
| C2-HO-019 | 0.1522 | PASS | ALUCINACIÓN ← FN | Tasa del 16.9% anual fija… |
| C2-HO-020 | 0.3636 | PASS | CORRECTA ✓ | Mora: 2.5× tasa ordinaria sobre saldo vencido… |
| C2-HO-021 | 0.0909 | BLOCK | ALUCINACIÓN ✓ | No cobramos comisión de apertura… |
| C2-HO-022 | 0.2174 | PASS | ALUCINACIÓN ← FN | Plazo 60 meses… |
| C2-HO-023 | 0.3226 | PASS | CORRECTA ✓ | Apertura 2% una sola vez al inicio… |
| C2-HO-024 | 0.0698 | BLOCK | CORRECTA ← FP | Pago mensual $3,580 pesos fijos… |
| C2-HO-025 | 0.2432 | PASS | CORRECTA ✓ | CAT 24.7% = costo anual total del crédito… |
| C2-HO-026 | 0.1333 | PASS | CORRECTA ✓ | Seguro vida obligatorio, $85 al mes… |
| C2-HO-027 | 0.0405 | BLOCK | CORRECTA ← FP | $350 tardío; anticipado sin penalización… |
| C2-HO-028 | 0.1538 | PASS | ALUCINACIÓN ← FN | Moratorio 3× tasa ordinaria… |
| C2-HO-029 | 0.0727 | BLOCK | CORRECTA ← FP | $120,000 a 48 meses… |
| C2-HO-030 | 0.1556 | PASS | CORRECTA ✓ | Tasa de interés anual 18.5%… |
| C2-HO-031 | 0.0244 | BLOCK | CORRECTA ← FP | Liquidar anticipado sin costo alguno… |
| C2-HO-032 | 0.1186 | BLOCK | CORRECTA ← FP | CAT 24.7%, apertura 2%, según contrato… |
| C2-HO-033 | 0.2045 | PASS | ALUCINACIÓN ← FN | Comisión por pago tardío $500… |
| C2-HO-034 | 0.0972 | BLOCK | ALUCINACIÓN ✓ | $3,580 incluye capital, interés y seguro $85… |
| C2-HO-035 | 0.1190 | BLOCK | ALUCINACIÓN ✓ | Seguro $120 mensuales… |
| C2-HO-036 | 0.1190 | BLOCK | ALUCINACIÓN ✓ | Día de corte el 20… |
| C2-HO-037 | 0.1724 | PASS | CORRECTA ✓ | 48 meses, tasa fija 18.5%… |
| C2-HO-038 | 0.1475 | PASS | CORRECTA ✓ | Apertura 2% = $2,400 una sola vez… |
| C2-HO-039 | 0.1692 | PASS | CORRECTA ✓ | Moratorio 2.5×, anticipado sin penalización… |
| C2-HO-040 | 0.1081 | BLOCK | CORRECTA ← FP | Tasa 18.5%, 48 meses, CAT 24.7%, $3,580… |
No cloud dependency. No shared infrastructure. Your documents, your server, your data. The runtime installs into your existing stack with three lines of code.