Ajah

Ajah is a self-hosted AI gateway that gives developers full control over their LLM integrations without relying on third-party services or exposing data externally. It acts as a proxy layer between your application and multiple LLM providers, routing requests while enforcing data privacy by keeping all traffic within your own server environment. Ajah's standout capability is real-time RAG verification — it checks AI-generated responses against source documents before returning them to the user, catching hallucinations at zero latency. For teams building complex multi-agent workflows, Ajah provides visual session tracing that makes it possible to debug and audit agent interactions step by step. It also automatically masks Personally Identifiable Information (PII) before any data reaches storage, helping teams meet compliance requirements without adding custom preprocessing logic. Detailed cost attribution per user or feature gives engineering leads clear visibility into LLM spend across their product. Setup is handled via Docker, making deployment accessible for any team with containerised infrastructure.

Used by developers and engineering teams to self-host an AI gateway with RAG verification, hallucination detection, PII masking, and multi-LLM cost tracking.