The simple version
What is kombify AI?
Imagine having a knowledgeable assistant that understands your homelab setup, answers questions about Docker and Linux, monitors your servers around the clock, and can even fix issues automatically. That is kombify AI.
Two types of intelligence
kombify AI is built around two distinct agent types that work together:Companions — Your conversational assistants
Companions are on-demand AI assistants with distinct personalities and persistent memory. They remember your preferences, understand your infrastructure, and help you get things done.Smart Home
IoT control and Home Assistant integration
Personal
Daily life, reminders, and context-aware assistance
Work
Productivity, projects, and team context
- Persistent memory across conversations
- Voice interaction via SpeechKit (6 STT providers, local TTS, Voice Agent Mode)
- Aware of your current infrastructure state
- Delegate infrastructure tasks to Managers
Managers — Your autonomous operators
Managers run 24/7 in the background, handling infrastructure operations without requiring your input.Guard
Autonomous watchdog — Monitors security and reliability across your infrastructure with three processing loops: fast (10s), heartbeat (60s), and deep scan (6h).
Server Manager
AI-powered sysadmin — Handles package management, user management, file operations, cron scheduling, and network configuration.
- Run autonomously, no manual interaction needed
- Trust progression model (Observer → Advisor → Autonomous)
- Event-driven, task-oriented operation
- Approval workflows for critical actions
How it works
You ask a question or issue arises
Either you start a conversation, or a Manager detects an event (e.g., a service goes down).
The Brain routes the request
The Brain classifies your intent and routes it to the right Companion or Manager. It is a stateless router, not an AI agent itself.
The agent acts
Companions respond conversationally with context from your infrastructure. Managers execute operations with appropriate approval.
Deployment options
- SaaS (Managed)
- Self-Hosted (BYOK)
Use kombify AI through kombify Cloud. Models are included — no API keys needed.
- Available at
chat.kombify.ioand embedded in the Cloud dashboard - Included in Cloud subscription plans
- No setup required
Get started with Cloud
Supported AI providers
kombify AI supports multiple model providers with intelligent routing:| Tier | Providers | Examples |
|---|---|---|
| Direct API | OpenAI, Anthropic, Google AI | GPT-4o, Claude Sonnet, Gemini |
| Cloud Platforms | Azure AI Foundry | Hosted model endpoints |
| Routers/Local | OpenRouter, Ollama | Any model, self-hosted LLMs |
Five frontend surfaces
Chat Widget
Embeddable floating bubble integrated into kombify Cloud
Web Chat
Full-featured chat at
chat.kombify.io with sidebar, sessions, and model pickerMobile App
Android-first with voice interaction and push notifications
Portal
Settings hub at
ai.kombify.io for companion configuration and key managementSpeechKit
Windows-first speech framework with STT/TTS and real-time Voice Agent Mode
Context-aware intelligence
kombify AI knows your infrastructure. When you ask “Why is my NAS slow?”, it can check:- Your active
kombination.yamlspecification - Current node status (CPU, RAM, disk usage)
- Recent error logs and service health
- Running containers and their resource consumption
Next steps
Quick start
Get started with kombify AI in 5 minutes
Set up Guard
Enable 24/7 autonomous monitoring
Architecture deep dive
Understand Brain, Companions, and Managers
Supported models
Browse supported AI providers and models