About

About SemanticGuard

What we build

SemanticGuard is an AI gateway with self-validating semantic caching for LLM API traffic. It sits in front of OpenAI, Anthropic, Google Vertex AI, Azure OpenAI, AWS Bedrock, and Mistral, caches semantically equivalent requests, and verifies every cache match before serving so customers never get the wrong response. The product surfaces real-time cost, savings, and request traces in a dashboard, exposes an OpenAI-compatible HTTP endpoint, and ships a one-line SDK integration via the @semanticguard/ai-sdk npm package.

Who we serve

Developers and small SaaS teams routing production LLM traffic who want to cut API costs without breaking responses. Free tier includes 10,000 requests per month and Shadow Mode for measuring potential savings before any cached response is served. Pro is $49 per month, Enterprise bills 15% of documented savings with a $500 per month minimum. No credit card required to start.

How we work

AI
We do not generate content. We cache responses from the upstream LLM you already use, and we make small auxiliary calls (template verification, intent extraction) on the cheapest model from the same vendor to keep cache matches correct. Auxiliary calls are billed to your account by the upstream provider, capped per tenant per hour to prevent runaway spend.
Integrations
OpenAI, Anthropic, Google Vertex AI, Azure OpenAI, AWS Bedrock, Mistral. Plus the Vercel AI SDK via withSemanticGuard(), an OpenAI-compatible HTTP base URL for any framework that accepts a custom endpoint, and a bundled MCP server for Claude, Cursor, and other AI dev tools.
Security
Data encrypted in transit (TLS 1.2+) and at rest. Neon Postgres for primary storage, Upstash Redis and Upstash Vector for cache state. JWT sessions with optional TOTP MFA and WebAuthn passkeys. PII redaction for API keys, tokens, emails, and phone numbers is on by default before any prompt content is written. Upstream API keys are passed through at request time and never stored in plaintext; only a one-way SHA-256 hash is retained.
Privacy
We do not sell data. We do not use your prompts, responses, or any user content to train machine learning models. Prompt and response storage is opt-in per tenant; when disabled, only request metadata is logged. Upstream LLM providers process requests under their own terms and cannot train on your data through us.

Company

SemanticGuard is a product of KB-AI LLC, a Wyoming limited liability company. KB-AI LLC also publishes other AI software. See www.kbaicorp.com.

KB-AI LLC
30 N Gould St Ste R
Sheridan, WY 82801
United States
Contact: support@semanticguard.dev

Resources