nasiko: a developer control plane for centralized management and intelligent routing of containerized AI agents

What it solves

Nasiko is a developer control plane designed to manage AI agents at scale. It solves the complexity of deploying, routing, and monitoring multiple specialized AI agents within a production environment, providing a centralized system for agent lifecycle management and intelligent traffic distribution.

How it works

Nasiko uses a cloud-native microservices architecture centered around the Kong API Gateway. It employs a centralized registry to store agent metadata via "AgentCards" (capability definitions). A LangChain-powered router analyzes user queries and semantically matches them to the most suitable agent based on these capabilities. Agents are deployed as containerized services (Docker/Kubernetes) and are automatically registered with the gateway for discovery and load balancing.

Who it’s for

It is built for AI developers and enterprise teams who need to orchestrate a network of specialized AI agents rather than relying on a single monolithic LLM application.

Highlights

• Intelligent Routing: Uses LangChain to analyze intent and route queries to the best agent based on confidence scoring.
• Agent Lifecycle Management: Supports version-controlled storage, hot deployment, and automated containerization.
• Enterprise Infrastructure: Integrates Kong API Gateway for routing, GitHub OAuth for authentication, and Arize Phoenix for LLM-native observability.
• Developer Tooling: Includes a dedicated CLI for agent uploads and a web-based dashboard for platform management.