ChatGPT, Claude, Gemini – we all use AI assistants in our daily lives now. But if you work with sensitive data or simply value your privacy, you might ask yourself: Do I really need to send my texts, code snippets, and documents to the cloud?
The answer: No. With OllamaDeploy, I've developed an open-source tool that automates the installation of a complete AI system on Mac – in less than 10 minutes.
What is OllamaDeploy?
OllamaDeploy is an installation script that brings together three components:
- Ollama – the foundation platform for local Large Language Models (LLMs)
- Docker Desktop – for container isolation
- Open WebUI – a ChatGPT-like web interface
After installation, a complete AI system runs locally on your Mac. No internet connection required, no data leaves your machine.
Why Local?
The advantages of a local AI installation are numerous:
- Privacy – Confidential documents, code, and personal data stay on your own device
- No subscription costs – Once installed, there are no monthly fees
- Offline operation – Perfect for travel or environments without internet access
- Flexibility – Multiple AI models can be used in parallel, depending on the task
Legal Considerations
If you use LLMs for business purposes, you should be aware of the legal framework. GDPR applies to local processing as well, but offers clear advantages over cloud services: There's no risk of data transfer to third countries, and responsibility lies solely with the operator – no complex data processing agreements required.
For purely private use, the GDPR's so-called household exemption applies. A comprehensive overview of the legal details can be found in the Heise article "Rechtliche Tipps zum Selfhosting von LLMs" (German).
Which Models are Available?
OllamaDeploy supports all models from the Ollama catalog:
- LLaMA 3.1/3.3 (Meta) – The all-rounder for most tasks
- DeepSeek-R1 – Particularly strong for programming tasks
- Gemma 3 (Google) – Compact and fast
- Mistral – Good balance of speed and quality
System Requirements
OllamaDeploy is optimized for Macs with Apple Silicon:
- Mac with M1, M2, M3, or M4 processor
- At least 16 GB RAM (32 GB recommended)
- 50 GB free storage space
- macOS 11 Big Sur or newer
Installation in 3 Steps
The installation is intentionally simple:
# 1. Download and extract ZIP
# Download: https://juergenkoller.software/downloads/OllamaDeploy.zip
cd OllamaDeploy
# 2. Start the interactive menu
./start.sh
# 3. Select "Fully Automatic Installation"After about 10 minutes, the web interface opens at http://localhost:3000.
Who is OllamaDeploy For?
The tool is aimed at:
- Developers – Code reviews, debugging, documentation
- Content Creators – Writing texts, brainstorming, translations
- Businesses – Processing sensitive data locally
- Private users – Using AI without data sharing
Download
OllamaDeploy is available as a free download. The project is based on proven open-source components (Ollama, Open WebUI) and can be freely used.
Download OllamaDeploy (ZIP, 45 KB)
Based on: heise.de – Mac als lokales KI-System: So geht's
Conclusion
If you want to use AI without sending your data to cloud services, OllamaDeploy offers a straightforward solution. The installation is automated, the web interface is intuitive, and the model selection covers most use cases.
Give it a try and let me know your feedback – via email.