Streamlit AI assistant — a documentation chatbot for Streamlit

Ever wanted to chat with the Streamlit documentation? Well, with the power of Snowflake Cortex now you can!

You can try out the AI Assistant app below:

Running it yourself

Snowflake backend setup

1. Open the worksheet at snowflake/initial_setup.sql in Snowflake and run it.

2. Open the notebook at snowflake/populate_st_assistant_data.ipynb in Snowflake and follow the instructions in there.

Try the app on your machine

1. Get the code:

   $ git clone https://github.com/streamlit/demo-ai-assistant

2. Start a virtual environment and get the dependencies (requires uv):

   $ uv venv
   $ .venv/bin/activate
   $ uv sync

3. Add your Snowflake account to .streamlit/secrets.toml under [connections.snowflake]

   [connections.snowflake]
   account = "YOUR_ACCOUNT_GOES_HERE" # <-- Change this
   # This is the user we set up for you in the pipeline setup steps:
   user = "ST_ASSISTANT_USER"
   # In the Snowflake UI, create an API token for the user above and paste here:
   password = "PASTE_TOKEN_HERE" # <-- Change this
   # Everything below was already set up in the pipeline steps:
   role = "ST_ASSISTANT_USER"
   warehouse = "COMPUTE_WH"
   database = "ST_ASSISTANT_DEV"
   schema = "PUBLIC"

4. Start the app:

   $ streamlit run streamlit_app.py