Open Source · PowerShell · Windows 11 · AI Stack
Your AI stack.
Your machine.
No cloud required.
A modular PowerShell toolkit that installs Ollama + Open-WebUI on Windows 11 in one script run. Privacy-first, fully offline, OpenAI-compatible API included.
The stack
# What you get after running the scripts
Windows 11
└─ Python 3.11+
└─ Open-WebUI http://localhost:3000
│ ChatGPT-like UI + OpenAI-compatible REST API
└─ Ollama http://localhost:11434
│ LLM runtime + model manager
└─ Local LLM e.g. qwen2.5:7b, mistral, llama3.2
stored locally — no network calls during inference
Windows 11
└─ Python 3.11+
└─ Open-WebUI http://localhost:3000
│ ChatGPT-like UI + OpenAI-compatible REST API
└─ Ollama http://localhost:11434
│ LLM runtime + model manager
└─ Local LLM e.g. qwen2.5:7b, mistral, llama3.2
stored locally — no network calls during inference
Phases
Phase 1 — Done
Local AI stack
- Docker audit + cleanup
- Ollama install + model download
- Open-WebUI install + configure
- Environment validation
- HTML health report
Phase 2 — Planned
Workflow automation
- n8n workflow engine
- Docker Compose stack
- AI ↔ calendar / email hooks
- MCP server integration
- Scheduled pipelines
Scripts
00-System-Check
Pre-flight — CPU, RAM, disk, WSL, Python version check
01-Docker-Audit
Full inventory of running containers, images, volumes — JSON report
02-Docker-Clean
Remove dangling images, stopped containers, unused volumes
03-Install-OpenWebUI
pip install open-webui, create startup scripts, configure port
06-Configure-OpenWebUI
Point Open-WebUI at Ollama, set model defaults, disable telemetry
07-Validate-Environment
End-to-end health check — API ping, model list, UI reachability
99-Generate-Report
HTML summary report with stack status, versions, memory usage
Quick start
01
Clone the repo and open PowerShell as Administrator
02
Run
00-System-Check.ps1 — confirms prerequisites are met
03
Run
03-Install-OpenWebUI.ps1 — installs Open-WebUI via pip
04
Install Ollama from
ollama.com, pull a model: ollama pull qwen2.5:7b
05
Run
06-Configure-OpenWebUI.ps1 and open http://localhost:3000