AI Agents in Action: Building Smart, Open-Source LLM-based Workflows

Welcome to the official repository for the AI Agents Workshop! This guide will walk you through setting up your environment and running the exercises.

Getting Started

Follow these steps to get everything up and running.

1. Install uv

This project uses uv, a fast Python package and project manager. If you don't have it installed, you can use pip:

pip install uv

For other installation methods, see the official uv documentation.

2. Install Dependencies

Once uv is installed, sync the project dependencies:

uv sync

This command installs all the necessary packages listed in the uv.lock file.

Running the Application

You have two options for running the application:

1. Using the Command Line: Run the following command in your terminal at the project root:

   uv run chainlit run aia25/app.py -w -h

2. Using the Shorthand Command: First, install the project in editable mode:

   uv pip install -e .

   Then, you can use the shorthand command:

   uv run app

   This will start the user interface.

Tracing

You can have a look at the flow of execution, by inspecting the traces. Traces are separated by experiment, and each time you start a specific exercise, a new trace is generated. We are using MLFlow for tracing, to launch the dashboard, just enter the following command in your CLI (ensure your virtual environment is activated):

 mlflow ui --backend-store-uri sqlite:///mlflow.db --port 5001

This starts the MLFlow server on port 5001. To inspect your server just go to: http://localhost:5001/

Repository Structure

The repository is organized into the following directories:

• aia25/: Contains the core application logic and user interface, powered by Chainlit.
• exercise01/ to exercise04/: These folders contain the exercises for the workshop.
• solution_exercise02/ to solution_exercise04/: These folders contain the solutions to the corresponding exercises.
• images/: Contains images used in the documentation.

How to Use the Workshop App

1. Start the application using one of the methods described above.

2. Select an agent to interact with by clicking the gear icon on the left side of the chat interface.

   

3. Choose the agent you want to run from the dropdown menu. Each agent corresponds to an execute_agent function within the my_agents.py file of each exercise module.

   

Now you're all set! Enjoy the workshop.