Decide - Data Analysis Assistant

Decide is a Streamlit-based data analysis assistant that lets you chat with your data using open-source LLMs via Ollama.

Quickstart

1. Requirements

• Python 3.9+
• Ollama running locally (for LLM inference)
• uv (Python dependency management)

2. Run LLM locally

Install Ollama from https://ollama.com/download.

Start Ollama and pull a model, e.g. gpt-oss:20b:

ollama pull gpt-oss:20b

Important: Make sure Ollama is running as a service:

ollama serve

Keep this terminal window open as Ollama needs to be running for the application to work.

3. Install Dependencies

Install the project dependencies using uv:

uv sync

4. Run the Application

You can run the application in two ways:

Option 1: Using the provided script

./scripts/run.sh

Option 2: Using uv directly

uv run streamlit run main.py

The application will start and be available at http://localhost:8501 in your browser.

Linting, formatting and type checking

This project uses ruff for litning and formatting and mypy for type checking. You can run all with:

./scripts/check.sh

Logging

This project uses Logfire for observability and monitoring.

Setup Logfire (Optional)

1. Get a Logfire token:

   • Sign up at https://logfire.pydantic.dev/
   • Create a new project and get your token

2. Configure the token: Create a .env file in the project root:

   echo "LOGFIRE_TOKEN=your_token_here" > .env

Viewing Logs

Once configured, you can view your logs and metrics at your Logfire dashboard URL, which will be displayed in the terminal when the application starts.