Night Planner Agent

The Night Planner Agent is an AI agent that helps users plan an evening out by suggesting dinner and event combinations based on user preferences. It uses the smolagents library along with custom tools - the Google Places API to find restaurants and the Ticketmaster API to find events.

Installation and Setup

Tested with Python 3.10. You need to have a Google Places API key and a Ticketmaster API key to use the event planner.

This project uses the Hugging Face inference endpoints with Qwen/Qwen2.5-Coder-32B-Instruct as the base model. smolagents allows replacing this with local transformers compatible models, ollama endpoint, closed sourced model endpoints, etc. See the smolagents docs for details.

Installation

Clone the repository and install the requirements.

git clone https://github.com/yourusername/night-planner-agent.git
cd night-planner-agent
pip install -r requirements.txt

Create an .env file in the root folder and copy paste the following with your own API keys.

GOOGLE_PLACES_API_KEY=your_google_places_api_key
TICKETMASTER_API_KEY=your_ticketmaster_api_key

Login to Hugging Face by entering your token, this is required to use the HF inference endpoints.

huggingface-cli login

Running the Code

You can run the Night Planner Agent in two modes: command-line mode and interface mode. To run the agent with a user query from the command line, use the following command:

python app.py --query "I want to plan a date night in New York City next Saturday. Looking for an Italian dinner and maybe a comedy show. Mid-range budget."

Alternatively, launch the Gradio demo to use the agent in interactive mode.

python app.py --interface

License

This project is licensed under the MIT License.