RAINavigator

Pre-model validation of user input: injection detection, topic blocking, encoding checks.

• practice-derived3 triggering use cases require Input Rails / Prompt Shields

Faithfulness scoring of answers against retrieved sources; deterministic fallback instead of hallucination; schema-validated structured output.

• practice-derived3 triggering use cases require Output Rails / Groundedness Check

Automated detection, pseudonymisation and blocking of personal data in inputs, retrievals and outputs.

• practice-derived3 triggering use cases require PII Scrubbing / DLP-NER Layer

Relevance, freshness and per-user permission checks on every retrieved chunk; curated, versioned index.

• practice-derived3 triggering use cases require Retrieval Rails (ACL-aware RAG)

Synthetic-content labelling & watermarking: visible disclosure plus machine-readable provenance (C2PA Content Credentials) embedded in generated images, audio and video; metadata identifying artificial origin survives common transformations. Discharges Art. 50(2)/(4) for deepfakes and synthetic media; verification telemetry (watermark presence/validity checks at publication gates) is the corresponding evidence stream.

• practice-derived3 triggering use cases require Synthetic-Content Labelling / Watermarking

Computes a probabilistic confidence score for every output and holds the transaction when the score falls below the workflow's regulatory threshold.

• practice-derived1 triggering use case require Confidence Scoring & Threshold Gate

Omnichannel voice/chat agent runtime with knowledge grounding, AI-disclosure prompts, consent handling and transcript capture.

• practice-derived1 triggering use case require Conversational Care Engine

HITL escalation queue & review UI ('Human-as-a-Tool': the agent calls the human like any other tool via propose-action objects). Confidence- and risk-threshold routing, SLA timers, structured accept/modify/reject verdicts with digital reviewer signature at gate release — each verdict is itself Art. 14 evidence and feeds the active-learning loop.

• practice-derived1 triggering use case require HITL Escalation Queue & Review UI

Operator stop controls and degraded-mode fallbacks; real-time override (veto) channels for HOTL operation.

• practice-derived1 triggering use case require Kill Switch / Graceful Degradation

Continuous evaluation against golden sets and sampled human verdicts; raises drift alerts and feeds the recertification cycle.

• practice-derived1 triggering use case require Model Drift & Accuracy Monitor

Vendor-neutral abstraction over 20+ foundation models across text, vision, audio and code; decouples application logic from any single provider's availability, pricing or deprecation cycle.

• practice-derived1 triggering use case require Multi-Model Orchestration Layer

Abstraction layer decoupling application logic from model providers: dynamic routing on capability, cost, latency SLA and regulatory constraint (sensitive-data classes pinned to ZDR private/VPC endpoints or on-prem open-weight instances); real-time health monitoring with automatic fallback to secondary endpoints or local fine-tuned models on outage/latency spikes. Discharges resilience duties (DORA-class), prevents provider lock-in, and makes model deprecations a routing-table change instead of a re-architecture. Router decisions are logged into the decision trace — model version per event is an audit-packet field.

• practice-derived1 triggering use case require Multi-Model Router & Fallback Abstraction