Audience
Enterprises interested in a solution to build AI apps and integrate AI agents securely into their systems
About Arch
Arch is an intelligent gateway designed to protect, observe, and personalize AI agents through seamless integration with your APIs. Built on Envoy Proxy, Arch offers secure handling, intelligent routing, robust observability, and integration with backend systems, all external to business logic. It features an out-of-process architecture compatible with various application languages, enabling quick deployment and transparent upgrades. Engineered with specialized sub-billion parameter Large Language Models (LLMs), Arch excels in critical prompt-related tasks such as function calling for API personalization, prompt guards to prevent toxic or jailbreak prompts, and intent-drift detection to enhance retrieval accuracy and response efficiency. Arch extends Envoy's cluster subsystem to manage upstream connections to LLMs, providing resilient AI application development. It also serves as an edge gateway for AI applications, offering TLS termination, rate limiting, and prompt-based routing.