Build an LLM Chat app in 5 minutes

Jump ahead:

• Quickly connect to Snowflake's LLM REST API
• Building your first app

Overview

This repository shows you how to build a simple chat app using Snowflake's API.

Quickly connect to Snowflake's LLM REST API

curl "https://[ACCOUNT-IDENTIFIER].snowflakecomputing.com/api/v2/cortex/inference:complete" \
    -H "Content-Type: application/json" \
    -H "Authorization: Bearer YOUR_PERSONAL_ACCESS_TOKEN" \
    -d '{
        "model": "claude-3-5-sonnet",
        "messages": [
            {
            "role": "user",
            "content": "Write me a one line poem about Snowflake"
            }
        ]
    }'

Snowflake's API is compatible with OpenAI's spec

To reproduce:

1. Sign up for a new Snowflake account here. Activate your account by e-mail (see screenshot below), pick a username and password, and you should see a welcome screen.

   <img width="300" src="https://github.com/user-attachments/assets/1a11cd4c-514d-4ac0-98b2-5e2207dc6698" />

   Note: if you can't find your login info, search your inbox for an e-mail from "Snowflake Computing".

2. Generate a Personal Access Token (PAT) -- this is your API key -- by visiting Authentication Settings. Be sure to set the network policy exception for easy weekend hacking:

   <img width="605" height="375" alt="Screenshot 2025-09-12 at 11 25 08 AM" src="https://github.com/user-attachments/assets/6ada1e18-4985-4bdd-a5a1-cbae7f78d8f9" />

3. Find your account identifier by clicking on your name in the bottom left corner of Snowflake's UI, and selecting Connect a tool to Snowflake:

   <img width="220" alt="Screenshot 2025-04-15 at 9 08 27 PM" src="https://github.com/user-attachments/assets/d485f628-e397-4869-82d7-c5e1c1af24c0" />

   <img width="450" src="https://github.com/user-attachments/assets/c4839cb9-585f-4af7-a7fb-40c328153786" />

4. That's it! Run the curl command in your terminal and observe the streaming output

Building your first app

1. Clone this repo.

2. Sign up for a new Snowflake account here. Activate your account by e-mail (see screenshot below), pick a username and password, and you should see a welcome screen.

   <img width="300" src="https://github.com/user-attachments/assets/1a11cd4c-514d-4ac0-98b2-5e2207dc6698" />

   Note: If you can't find your login info, search your inbox for an e-mail from "Snowflake Computing".

3. Generate a Personal Access Token (PAT) -- this is your API key -- by visiting Authentication Settings. Be sure to set the network policy exception for easy weekend hacking:

   <img width="605" height="375" alt="Screenshot 2025-09-12 at 11 25 08 AM" src="https://github.com/user-attachments/assets/6ada1e18-4985-4bdd-a5a1-cbae7f78d8f9" />

4. Click on your name in the bottom left corner, and select Connect a tool to Snowflake:

   <img width="220" alt="Screenshot 2025-04-15 at 9 08 27 PM" src="https://github.com/user-attachments/assets/d485f628-e397-4869-82d7-c5e1c1af24c0" />

5. Use the information in the connect dialog to fill in your secrets.toml file:

   <img width="450" src="https://github.com/user-attachments/assets/c4839cb9-585f-4af7-a7fb-40c328153786" />

    [snowflake]
    account = "[Account Identifier]"
    user = "[User Name]"
    api_key = "[Paste in your Personal Access Token here]"
    role = "[Role]"
    host = "[Account/Server URL]"

6. Run pip install -r requirements.txt to make sure you have all the dependencies working.

7. Run streamlit run streamlit.app and you should see a simple chat app ready to chat with you! Try changing the model used: mistral-large2, claude-3-5-sonnet or llama3.1-70b and compare the chat interaction!