Lang.ai Native Snowflake application provides AI agents for product managers that leverage your existing Snowflake data (unstructured and structured data) to serve meaningful product recommendations —with context— directly in Slack.
This quickstart is a guide to installing and running your first Lang AI agent for data analysis in your own Snowflake infrastructure.
Prerequisites
- Being part of Lang.ai AI Agent Design Partner program
- Sharing with Lang.ai's Account Executive your Snowflake Account Identifier
What You'll Learn
- Installing and setting up the Lang.ai Native Application
- Creating and running your first AI agent
What You'll Need
- A Snowflake account with ACCOUNTADMIN access to grant account level privileges, allow external connections and create a Snowflake View.
What You'll Build
- A Snowflake AI Agent for data analysis
In this step we will be creating a sample database and a sample view. We will be using this data in the following steps to create our first AI agent.
Sign in to your Snowflake account and create a new SQL Worksheet.
To start, copy the
This script will:
- Create a Database called zoom_cancellations and populate it with demo data
- Create a Database called zoom_users and populate it with demo data
- Create a View called zoom_cancellations_view to connect the data from both tables
After the script has successfully executed, you will see a LANG_AI_DEMO database with the following tables and one view:
The view created by the script already includes the required fields to create AI agents:
In this example, we are creating a simple view with the fields id, text, creation_date, user_id, and Customer_Spend.
The last column Customer_Spend is not mandatory, but it is included as we will configure the agent to use it to group the insights by the customer monthly spending.
You may include additional columns that may be used by the AI agent to aggregate the insights generated.
Log in to Snowsight as an ACCOUNTADMIN and follow these steps to install the Lang.ai Native App:
1. Install the application
Go to Data Products > Private Sharing > Shared With You and click on Get to install the app.
Once installed, click on the name of the application to open the app.
2. Grant privileges
Click Grant to grant the application the necessary privileges:
Account-level Privileges
- The BIND SERVICE ENDPOINT privilege enables the services in the app to connect to each other.
- The CREATE WAREHOUSE and CREATE COMPUTE POOL are required by the Lang.ai Native App to create and manage resources for performing service operations. Learn more [here]((https://help.lang.ai/en/articles/9813363-install-the-native-app-with-an-sql-script).
Give access to the Snowflake View
Share data with your AI Agents by creating and sharing access to a view. Learn more here.
-- Customize if needed
SET LANGAI_APP_NAME = 'LANGAI_APP';
--- Give the application access to the view
GRANT USAGE ON DATABASE "lang_ai_demo" TO APPLICATION IDENTIFIER($LANGAI_APP_NAME);
GRANT USAGE ON SCHEMA "PUBLIC" TO APPLICATION IDENTIFIER($LANGAI_APP_NAME);
GRANT SELECT ON VIEW "zoom_cancellations" TO APPLICATION IDENTIFIER($LANGAI_APP_NAME);
3. Launch the app
Click Activate to activate the application.
Navigate back to the Data Products > Apps page and wait for the spinner in the INSTALLED column to stop running. When it's done you'll see "1 minute ago" in that column. Then click Launch App to start the application.
Learn to run your fist AI agent on top of Snowflake data.
Select Your Goal
Begin by selecting the goal you want to achieve with your AI agent.
Select Your AI Agent
Choose the AI agent that aligns with your goal. Once selected, you will be prompted to choose the attributes that will be used to segment the insights. For example, selecting "subscription plan" will ensure insights are generated with this in mind, helping the AI agent identify patterns relevant to your business objectives.
Run Your AI Agent for the First Time
After completing the setup and creating your first agent, you're ready to run it for the first time.
To manually run your agent, click on Run Agent. The demo data has comments from July 2024, so make sure to select the entire month.
Navigate Your Agent Insights
Once the agent has finished processing, click on table to see the generated insights. In the insights view, you can navigate through the data by sorting and searching.
Review the Data Associated with the Insights
To view the unstructured text that was analyzed to generate an insight, simply click on the specific insight.
Additionally, you have the option to download the data from your active filters by clicking on Download CSV.
By following this guide, you have successfully set up an AI Agent running on Snowflake to automate your data analysis tasks.
Related Resources
What You Learned
- How to set up the Lang.ai Snowflake Native App
- How to create and configure an AI Agent in Lang.ai