Chatbot-with-Streamlit-Snowpark-OpenAPI

Overview

A demo of chatbot that prompts using natural language to explore data interactively.

Because chatbot will run in local machine using Streamlit, user will need to complete the preparation below.

Prerequisites

• An API key for OpenAI or another Large Language Model
• A service account to Snowflake environment
• Ability to install software on local machine such as minconda, streamlit
• (Optional) VSCode or the IDE of your choice installed

Preparing local machine

Complete the following steps in your local machine

1. Install Miniconda. This provides an isolated local environment for running Python anv various packages from Anaconda repository.

2. If you're using a machine with an Apple M1 chip, run the following command to use conda to create a Python 3.10 virtual environment, add the Snowflake conda channel, and install the numpy and pandas packages:

   conda create --name py310_env --override-channels -c https://repo.anaconda.com/pkgs/snowflake python=3.10 numpy pandas

3. Activate the environment created in those instructions by running conda activate py310_env and proceed to step 6 below.If you're not using a machine with an Apple M1 chip, continue to step 4.

4. Create a conda environment by running the following command:

   conda create --name snowpark -c https://repo.anaconda.com/pkgs/snowflake python=3.10

5. Activate the conda environment by running the following command:

   conda activate snowpark

6. Install Snowpark for Python, Streamlit, and OpenAI by running the following command:

   conda install -c https://repo.anaconda.com/pkgs/snowflake snowflake-snowpark-python openai streamlit

Set up workspace

1. Select a local directory to clone Chatbot With Streamlit Snowpark OpenAPI (ex. xkxc247\workspakce)

2. Clone using following command

   gh repo clone khanh-chau_wwg/chatbot-with-streamlit-snowpark-openapi

3. Open Terminal and navigate to chatbot-with-streamlit-snowpark-openapi directory

4. Use editor to create andup .streamlit\secrets.toml with the following info

   # .streamlit/secrets.toml # provide your own api key is optional

   OPENAI_API_KEY = ""

   # use the ATLAN_SVC only for testing. This user and following info has limited access to data in Snowflake Dev [connections.snowpark]

   user = "ATLAN_SVC"

   password = "Graing3r"

   warehouse = "DMG_WH_M"

   role = "ATLAN_SVC"

   account = "wwgraingerdev.us-east-1"

5. Run command

   streamlit run validate_credentials.py

If you encounter errors, please ensure .streamlit\secrets.toml has the correct information.

Create API Key from OpenAPI

Follow instructions to create OpenAPI API Key (https://www.howtogeek.com/885918/how-to-get-an-openai-api-key/)

Run Chatbot using sample customer demographics data

1. Open Terminal and navigate to chatbot-with-streamlit-snowpark-openapi directory

2. Ensure .streamlit\secrets.toml uses the ATLAN_SVC user information for Snowflake

3. Ensure snowflake.chatbot.py has the following line uncommented

   from customer_prompts import get_system_prompt

4. Run command

   streamlit run snowflake_chatbot.py

5. If all goes well, chatbot will introduces itself as Frosty

Hello there! I'm Frosty the Snowman, and I'm here to help you with your queries about the DMG.ML_TEST.CUSTOMER_INFO_VW table.

...

Try the following prompts

Can you provide the customer ID, gender, and income of customers from

California? Limit to top 10 highest income.

Please add median income for both lower and upper bound to result?

Let's limit result to Fremont city in California.

Would you add adjacent cities of Fremont to result, also add city name

Can you provide the demographics of customers by gender and income for customers from California?

Create your own data set

The sample code provide 3 different Snowflake data sets for chatbot to use: cybersyn_prompts, callcenter_prompts and customer_prompts. The default uses customer_prompts.py. By uncommenting the appropriate code, chatbot will use the

#from callcenter_prompts import get_system_prompt

#from cybersyn_prompts import get_system_prompt

from customer_prompts import get_system_prompt

To create your own data copy the customer_prompts.py and rename to something relevant to your data set.

Then make the following changes

• Set QUALIFIED_TABLE_NAME = "<database.schema.view>" to the Snowflake table or view that you want touse
• Set TABLE_DESCRIPTION = "" with hints and rules of your data. See examples in the customer_prompts.py
• Be sure to modify .streamlit/secrets.toml to use the Snowflake credential that has access to the data

Troubleshoot errors

• Be sure to connect to Global Connect as Snowflake environment uses IP whitelisting
• Ability to connect to OpenAPI requires valid API key. Also OpenAPI charges for usage so be sure there's enough credits to power LLM interactions.