The Scenario
Three analysts just joined the team. They have Snowflake access and a list of 20 core tables they'll be querying every day. What they don't have is any documentation of what's in those tables — column names, data types, which fields are nullable, what the default values are.
You're the data engineer responsible for onboarding them. You built most of these tables two years ago, so you know the schemas. But writing a data dictionary for 20 tables by hand, across potentially 300+ columns, is a week's worth of work you were not allocated.
The bad version:
- Run DESCRIBE TABLE for each of the 20 tables in Snowflake one at a time, copying the output from the results grid into the sheet.
- Manually add column headers, reformat the nullable column from Y/N to TRUE/FALSE for consistency, and merge in any default values that didn't show up in the DESCRIBE output.
- Get to table 14, realize you mis-labeled a column's data type in table 8, go back and fix it, and lose another 20 minutes.
Twenty tables is not a large number, but 300 columns is, and the moment someone adds a column to a table your manually maintained dictionary is already wrong.
The Easy Way: One Prompt in SheetXAI
SheetXAI is an AI agent that lives inside your Google Sheet. It can describe every column in a Snowflake table — name, type, nullability, default — and write the results directly into a sheet. Run it across a list of table names from column A and it iterates through all 20 without you touching anything.
Paste this into the SheetXAI sidebar:
For each table name listed in column A of this sheet (20 rows), describe its columns in Snowflake database PROD schema PUBLIC and write the results into this sheet as additional rows with columns: table_name, column_name, data_type, nullable, default_value
What You Get
- One row per column across all 20 tables, with
table_namerepeated in column A for every column of that table column_name,data_type,nullable(TRUE/FALSE), anddefault_valuein columns B–E- Tables processed in the order they appear in column A
- If a table doesn't exist in the schema, a single row appears for that table with an error note in column B
What If the Data Is Not Quite Ready
You want to add a primary key indicator column
For each table in column A, describe its columns in Snowflake PROD.PUBLIC and write table_name, column_name, data_type, nullable, default_value, and whether the column is part of a primary key constraint into columns A–F of this sheet
You want to filter the output to only non-nullable columns
For each table in column A (20 rows), describe its columns in Snowflake PROD.PUBLIC — write only the non-nullable columns into this sheet with columns table_name, column_name, and data_type
You need to document a single large table with 80+ columns
Describe all columns in Snowflake table PROD.PUBLIC.EVENTS and write column_name, data_type, kind, nullable, and default_value into columns A–E of this sheet, starting at A1 with headers
You need to catch columns that changed type since last documented and highlight them
Look at the table_name and column_name combinations listed in columns A and B of the Existing Tab, then describe those columns in Snowflake PROD.PUBLIC — write the current data_type in column C and flag 'TYPE CHANGED' in column D if it doesn't match what's in column C of the Existing Tab, then do a fresh full describe and append any new columns not in the list at the bottom with 'NEW' in column D
All four steps — compare, flag, append, label — in a single ask.
Try It
Get the 7-day free trial of SheetXAI and open the Google Sheet where your table list lives, then ask it to describe all 20 tables and write the full column inventory into the sheet. You can also explore profiling column value distributions or return to the Snowflake overview.