Use Cases
AI Gateway solves a handful of recurring problems for teams connecting applications to LLMs and AI agents to tools. They are grouped by the gateway they build on: the LLM Gateway or the MCP Gateway.
LLM Gateway
Unified LLM API
Reach any LLM provider from the SDK an application already uses, without rewriting it per provider.
Model Routing Across Deployments
Send each model to the provider, account, or deployment that serves it, from one endpoint.
Model Fallback
Keep LLM apps answering when the deployment a model is routed to can't be reached.
LLM Guardrails
Keep secrets, personal data, and jailbreak attempts out of LLM prompts and replies, enforced at the gateway.
Bring Your Own Key
Forward each caller's own provider API key upstream, without storing provider keys in the gateway.
LLM Token Usage Visibility
See token usage for every LLM exchange across providers, exported to your monitoring stack.
OpenAI Apps on AWS Bedrock
Serve applications written for the OpenAI API from models on Amazon Bedrock, using IAM credentials.
MCP Gateway
Turning Sources into Tools
HTTP APIs as Tools
Expose a plain REST API as MCP tools with no MCP server standing between it and the gateway.
OpenAPI Specs as Tools
Derive MCP tools and resources automatically from an existing OpenAPI document instead of hand-authoring each one.
Streaming Data as Tools
Turn a Kafka topic into MCP tools by chaining AI Gateway in front of a Kafka-backed REST API, with no custom MCP server and no application code.
Access Control
Centralized Auth
Enforce one auth policy at the gateway for every upstream MCP server and HTTP API, instead of duplicating auth logic per tool provider.
Secure Agent Access
Give an AI agent one authenticated endpoint to hold a credential for, no matter how many upstream tool providers sit behind it.