Query Snowflake data in Glean Assistant
Glean Assistant allows you to query your Snowflake data warehouse using natural language or direct SQL. This integration provides a governed, self-serve way to analyze approved datasets without leaving your chat interface or needing to write complex code.
By connecting Snowflake to Assistant, you can bypass the need to log into Snowsight for routine data checks. Assistant translates your questions into queries, fetches the results, and displays them in a structured table directly in your conversation.
Your organization controls which Snowflake projects and datasets are exposed to Glean Assistant. All queries are read-only, meaning Assistant cannot modify, delete, or update any data within your Snowflake warehouse.
This feature is currently in Beta. If you do not see Snowflake options in your Assistant, contact your internal Glean administrator to request access.
What you can do
- Natural Language Queries: Ask questions as if you were talking to a data analyst.
- Direct SQL execution: Run read-only SQL queries for precise, ad-hoc analysis.
- Tabular visualizations: View results in a clean, easy-to-read table format.
- Transparent citations: See exactly where the data came from and receive clear feedback if a query needs adjustment.
Query Snowflake data in Assistant
Ask questions in natural language
Use this method for general data exploration. Assistant will interpret your intent and map it to the relevant Snowflake tables.
Example prompts:
- "What was our quarterly revenue by region?"
- "Show the top 10 customers by usage growth this month."
- "Which support queues had the highest ticket volume last week?"
Run SQL queries directly
If you have a specific query in mind or need advanced filtering, you can input SQL commands directly into the chat.
Example:
SELECT region, SUM(arr) AS total_arr
FROM sales_summary
GROUP BY region
ORDER BY total_arr DESC;
Tips for better results
To get the most accurate answers from Assistant, keep these best practices in mind:
- Be specific: Include the exact metric like "ARR", dimension ("Region"), and time range ("Q3") in your prompt.
- Mention datasets: If you know the specific view or table name for example
SALES_SUMMARY, mention it to help Assistant find the right data faster. - Refine and rephrase: If the first answer seems incomplete, try rephrasing your question or switching to SQL for more granular control.
- Check permissions: Your ability to query data depends on your Snowflake role. If a specific table is not appearing, verify with your admin that you have
SELECTaccess.