Setup

Time: 15 minutes (do it before the event if you can)

You need two things: a model to talk to, and the workshop code. The default path runs a free, open model on your laptop with Ollama. If you would rather use a cloud provider, see Use any model provider.

Do this before the event

The model download is about 2.5 GB. Venue wifi with 40 people downloading at once is slow. Please set up at home if you can.

1. Install Ollama and a model

1. Install Ollama.

   macOS

   Download the app from ollama.com/download, or use Homebrew:

   brew install ollama

   Windows

   Download and run the installer from ollama.com/download.

   Linux

   curl -fsSL https://ollama.com/install.sh | sh

2. Download the workshop model. It is small, fast and good at calling tools.

   ollama pull qwen3:4b-instruct

   If your laptop has 8 GB of RAM or less, pick a smaller model instead: ollama pull qwen2.5:1.5b. It is less accurate, which actually makes the evaluation and attack modules more interesting.

3. Say hello.

   ollama run qwen3:4b-instruct "Say hello in five words"

   If you get a reply, Ollama is working. It runs in the background and listens on http://localhost:11434.

2. Get the code

1. Clone the repository and go to the code folder.

   git clone https://github.com/codewithsimon/ai-agent-workshop.git
   cd ai-agent-workshop/code

   No git? Download the ZIP from the repository page (green Code button, then Download ZIP).

2. Create a virtual environment and install the workshop package.

   macOS

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -e .

   Windows

   python -m venv .venv
   .venv\Scripts\activate
   pip install -e .

   Linux

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -e .

   The only dependencies are httpx (for HTTP requests) and python-dotenv (for settings). No AI vendor SDK.

3. Create your settings file.

   macOS

   cp .env.example .env

   Windows

   copy .env.example .env

   Linux

   cp .env.example .env

   The defaults point at Ollama on your laptop:

   code/.env

   LLM_BASE_URL=http://localhost:11434/v1
   LLM_API_KEY=ollama
   LLM_MODEL=qwen3:4b-instruct

3. Check your setup

From the code folder, run:

python -m agentlab.check

You should see something like:

Endpoint: http://localhost:11434/v1
Model:    qwen3:4b-instruct

1/2 Basic chat ... ok (2.1s): 'ready'
2/2 Tool calling ... ok (1.8s): model called get_session

You're ready for the workshop!

The first call can take longer while the model loads into memory. If something fails, see Troubleshooting.

Run all commands from the code folder

Every command in this workshop assumes you are in the code folder with your virtual environment active.

No luck installing? Use Codespaces

The repository includes a dev container. On GitHub, click Code, then Codespaces, then Create codespace. Python and the workshop package are installed for you. Codespaces has no GPU, so point .env at a cloud provider instead of Ollama. See Use any model provider.