Nasiko is a control plane that manages the lifecycle, routing, and observability of AI agents at scale. It provides centralized deployment, API gateway integration, and monitoring for production AI agent ecosystems.
Project overview
It unifies deployment, routing, observability, and lifecycle management into one control plane for AI agents.
Project type
AI Agent · Model Runtime · Infrastructure
Use cases
Automation
Deployment
Refer to project documentation
License
License pending
Best for
Developers, AI engineers, and operations teams managing production AI agent ecosystems.
Key capabilities
Provides a centralized registry, version-controlled storage, Docker-based automated deployment, and hot updates for agents.
Utilizes AI-driven query analysis, capability matching, and confidence scoring for agent selection.
Manages enterprise-grade APIs with plugins, load balancing, and automatic service discovery via Kong.
Maintains complete chat history, interaction tracking, and data persistence.
Provides a browser-based interface for platform management accessible via Kong Gateway.
Offers a command-line management interface for agents and platform status.
Provides a native desktop application for user interaction.
Features built-in monitoring and request tracing across microservices.
Limitations and risks
Requires Docker, Docker Compose, and Python 3.12+.
Generated superuser credentials and default JWT secrets need changing before production deployment.
Getting started
Setup requires cloning the repository, creating an environment configuration, installing Python dependencies using uv sync, running docker compose up -d, and accessing the web interface at http://localhost:9100/app/.
Alternatives and comparisons
Provides an AI-native proxy server and data plane that centralizes orchestration, safety, observability, and smart LLM routing.
Handles downloading, caching, GPU scheduling, and scaling for LLMs on Kubernetes.
Provides an open-source AI backend control plane that combines multi-model routing, an agentic tool harness, and isolated containerized workspaces.