FREE VERSION · SAFE · PRIVATE

Train your AI agents on your own device.

This public page is a guide only. Download the ZIP to enter the Lab.

Download AI-Agents-Lab-Windows.zip (v0.1.24) Simple Guide

LEARNING SPACE

University room

A professor and student agents turn a topic into a practical lesson, exercise, alternative solution, and feedback.

DISCUSSION SPACE

Conference room

A planner, researcher, and reviewer examine any topic with goals, alternative perspectives, and a practical next step in mind.

SERVICE BOUNDARY

Local by design

The Lab connects to Ollama, LM Studio, or vLLM running locally. It has no hosted AI API, analytics, or cloud chat backend.

LOCAL INFERENCE ENGINES

Choose the provider that fits your setup.

RECOMMENDED START

Ollama

A straightforward local runtime for Windows, macOS, and Linux.

Download Ollama

DESKTOP APP

LM Studio

A desktop app for running local models and an OpenAI-compatible local server.

Download LM Studio

HIGH-PERFORMANCE SERVING

vLLM

For Linux or compatible WSL2/GPU setups. Follow its installation guide; it is not a native Windows desktop download.

Open vLLM install guide

GET STARTED

Run the Lab locally.

  1. Download and extract the ZIP. Extract it to a folder you control, such as Documents\\AI-Agents-Lab. There is no hidden installation folder: this extracted folder is the Lab.
  2. Install Python and choose a local engine. Install Python 3.9 or later for Windows, then use one of the official provider links above.
  3. Run a local model or server. With Ollama, for example, run ollama pull qwen3:4b; the Lab uses its local address, 127.0.0.1:11434. LM Studio and vLLM expose OpenAI-compatible local servers on their configured loopback addresses.
  4. Keep Ollama local-only if you use it. Add "disable_ollama_cloud": true to %USERPROFILE%\.ollama\server.json, preserving any existing settings, then restart Ollama. Use downloaded local models, not names ending in :cloud.
  5. Start the local server from the extracted ZIP folder. Open start-windows.bat in that extracted folder (or run py -3 server.py there) and keep its window open. Its first run creates a Desktop shortcut that continues to open this extracted copy.
  6. Only after the local server has started, open the local address. When the start-windows.bat window confirms that the Lab is running, select Open Local Lab. The link cannot start the Lab by itself.

Manual updates — important: You do not download the ZIP each time you start the Lab. Download a new AI-Agents-Lab-Windows.zip only when you choose to update to a newer version. Keep your existing local models and back up data/campus.sqlite3 before moving to the new extracted folder. Then run start-windows.bat there once to update the Desktop shortcut.

PRIVACY

This site does not run the Lab.

There is no sign-in, chat input, model endpoint, telemetry, or personal-data collection on this Pages site. The download is a versioned local package; to use the Lab, extract it and connect only to a local Ollama, LM Studio, or vLLM installation.