Overview

Menatic AI is a self-hosted, Kubernetes-native platform that brings AI models, interactive notebooks, and intelligent automation to your infrastructure.

Key Features

Feature	Description
Dashboard	Real-time cluster overview with CPU, memory, and GPU metrics per node
AI Models	Deploy Ollama models locally or connect remote inference providers
Jupyter Notebooks	One-click notebook deployment with token-secured auto-login URLs
MCP Servers	Host Model Context Protocol servers as Kubernetes Deployments
Workflows & Builds	Container image build pipelines and automated workflow execution
RAG	Upload documents (PDF, text) and ground LLM responses in your own knowledge base
Admin Chat	AI-powered assistant with full access to cluster management tools
RBAC	Role-based access control with object-level permissions

Architecture

Menatic AI runs entirely on Kubernetes. All workloads — notebooks, MCP servers, and inference engines — are deployed as standard Kubernetes resources in a dedicated namespace. The platform backend is a Go HTTP server that interfaces with the Kubernetes API, while the frontend is a Next.js application.

┌─────────────────────────────────────────────────────────┐
│                    Kubernetes Cluster                    │
│                                                         │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────┐ │
│  │  ftn-ai ns  │  │   apps ns   │  │  kube-system ns │ │
│  │             │  │             │  │                 │ │
│  │  backend    │  │  notebooks  │  │  metrics-server │ │
│  │  frontend   │  │  mcp-servers│  │  ollama         │ │
│  │             │  │  workflows  │  │                 │ │
│  └─────────────┘  └─────────────┘  └─────────────────┘ │
└─────────────────────────────────────────────────────────┘

Quick Start

1. Follow the Deployment Guide to install on Kubernetes
2. Log in with the default admin credentials
3. Navigate to AI Models to deploy your first local model
4. Open Jupyter Notebooks to launch an interactive environment
5. Explore the Admin Chat to manage your cluster via natural language

Requirements

• Kubernetes 1.28+
• kubectl configured with cluster access
• Helm 3.x (for optional chart-based installs)
• metrics-server installed (for CPU/memory monitoring)
• NVIDIA GPU operator (optional, for GPU-accelerated inference)