PromptClassify deploys as a single container into your environment — detecting PII, PHI, and sensitive data in real time, scoring risk, and routing with confidence. Your rules, your infrastructure, your control.
Every prompt passes through a policy-driven classification pipeline — from ingestion to routing decision — in under 200 milliseconds.
Every inbound prompt is analyzed against your policy ruleset — combining pattern matching with entity recognition to detect PII, PHI, and sensitive data.
Detected entities are weighted by category, severity, and context to produce a normalized 0–10 risk score with full confidence metrics.
Policy-driven routing rules determine the action — route internally, sanitize and forward, block for review, or pass through — all based on your thresholds.
Every classification, score, and routing decision generates a traceable audit record with a unique ID — ready for your SIEM, observability stack, or compliance backend.
Every classification returns a rich, structured payload designed for immediate programmatic action and long-term auditability.
A normalized 0–10 risk score with a categorical level (LOW, MEDIUM, HIGH, CRITICAL) — immediately actionable for gateway routing or alerting.
Boolean flags indicating the presence of Protected Health Information and Personally Identifiable Information — enabling instant routing logic.
An explicit action (route internal, sanitize, block, pass), target, and human-readable reason — ready to drive your API gateway or workflow engine.
A unique audit ID, policy version, applied policy layers, and full classification metadata — immutable evidence for compliance and forensics.
Automatic tagging of applicable regulatory frameworks based on detected data types — HIPAA, GDPR, NIST, and more — surfaced in every response.
PromptClassify ships as a single Docker image. Deploy it where your data lives — no external calls, no data leaving your perimeter.
One Docker image. Pull, configure, run. Deploys into any container orchestration platform — Kubernetes, ECS, Docker Compose, or bare metal. Production-ready in minutes.
Run one instance or fifty. Stateless by design, PromptClassify scales horizontally behind your load balancer to meet any throughput requirement. Managed entirely by your team.
Structured stdout logs and native OpenTelemetry metrics, traces, and logs. Route to your SIEM, your observability backend, your audit system — through the mechanisms you already use.
PromptClassify maps detection categories and audit output directly to the frameworks your compliance team cares about.
PHI detection and routing controls designed for healthcare environments — ensuring patient data never reaches unauthorized endpoints.
Risk scoring and audit trails aligned with the NIST AI Risk Management Framework — providing measurable, documented governance controls.
Classification and policy controls that map to ISO 42001 requirements for establishing and maintaining an AI management system.
PII detection and data minimization support for EU data protection requirements — identify and control personal data before it reaches LLM providers.
Prompt injection detection, sensitive information disclosure prevention, and input validation aligned with OWASP's LLM-specific security framework.
Continuous policy updates from the PromptClassify cloud extend coverage as new standards emerge and existing frameworks evolve.
PromptClassify is the enforcement engine for AI GovOps — Amalfi AI's pioneering methodology that treats governance as a continuous operational flow embedded directly in infrastructure, not a periodic compliance exercise.
Learn more at Amalfi.AI →Every prompt is classified in real time — not sampled, not batched, not reviewed after the fact. Governance runs at the speed of inference.
Your governance rules are defined in structured policy layers — versioned, auditable, and enforceable. Compliance becomes an engineering artifact.
Every classification, score, and routing decision is logged with a unique audit ID. Evidence generation is automatic, not reconstructed.
Your PromptClassify instance automatically fetches updated rulesets from the PromptClassify cloud — ensuring your classification policies evolve as threats and standards change.
On startup and periodically during operation, your instance checks for new and updated classification rules from promptclassify.ai — no manual intervention, no downtime.
Cloud rulesets merge with your custom policies — your organization-specific rules always take priority. Add, override, or extend any classification pattern to match your environment.
Every ruleset update is versioned and recorded in the audit trail. You always know which policy version was active when a classification was made.
Request a demo to see how PromptClassify classifies, scores, and routes prompts in real time — deployed in your environment, governed by your rules.
Request a Demo →