Features

Core Capabilities

EdgeMask provides three core pillars of protection: security through automated redaction, cost optimization through semantic caching, and operational control through rate limiting.

Automated PII/PHI Redaction

Security

EdgeMask identifies and masks over 50+ sensitive entities at the edge before the prompt ever hits OpenAI's servers.

Supported Features

Social Security Numbers (SSN)

Credit Card numbers

Medical IDs & Patient records

API Keys & Secret tokens

Email addresses & Phone numbers

Custom entity patterns via regex rules

Pro Tip: All redaction happens at the edge node level — your sensitive data never leaves your network perimeter in plaintext.

Semantic Caching

Cost Reduction

Don't pay for the same query twice. EdgeMask uses vector-based semantic caching to identify similar prompts and serve cached responses instantly.

Supported Features

Save up to 40% on your monthly LLM bill

Reduce latency to <10ms for cached hits

Context-aware intent matching

Automatic cache invalidation policies

Per-model cache partitioning

Global edge cache distribution

Pro Tip: Semantic caching goes beyond exact-match. It understands intent, so paraphrased questions are served from cache too.

Rate-Limit Protection

Operations

Prevent individual users from draining your API credits with customizable per-user/per-key rate limits enforced at the gateway.

Supported Features

Per-user rate limit enforcement

Per-API-key budget controls

Sliding window & token bucket algorithms

Real-time usage dashboards

Custom alert thresholds

Automatic throttling & graceful degradation

Pro Tip: Rate limits are enforced at the gateway layer, meaning abusive requests never reach the provider — saving you money and headaches.