Databases & CRM
Set up Snowflake
Connect a Snowflake database. The same treatment as the SQL source — the schema is read, tables that hold your customers, orders and invoices become records with their figures totalled exactly — over Snowflake's own SQL API with a programmatic access token, so there is no driver, no app to register, nothing installed. Read-only.
Before you start
- A programmatic access token — in Snowsight, your name (bottom left) → Settings → Authentication → Programmatic access tokens → Generate — best issued for a user whose role is read-only on the database.
- The account identifier from your Snowsight address (the part before .snowflakecomputing.com, e.g. myorg-myaccount) and the database name.
- Only SELECT and SHOW statements are ever submitted, whatever else the token's role is allowed to do.
- Documents → Snowflake: enter the Account identifier and Database, and paste the token.
- Schema defaults to PUBLIC; set it only if your tables live in their own schema. Warehouse defaults to the token's own default — name one (an X-Small is plenty) only if that user has none.
- Optionally narrow with Only these tables or Skip these tables.
- Choose access groups, Test, then Sync.
What comes in
- Every table as a searchable entry — its columns, types and sample values, plus what it links to.
- Tables that look like registries also become one record per row, with its parent shown and a summary of the child facts that reference it (“orders: 40 · total amount 18,500”).
| Schema | The schema inside the database to read — blank uses PUBLIC. |
|---|---|
| Warehouse | The virtual warehouse queries run on — blank uses the token's own default; name one only if that user has none. An X-Small is enough for the modest, capped amounts this connector reads. |
| Only these tables | Allow-list, one table per line; blank reads every table. |
| Skip these tables | The opposite — read everything except these. |
Access
Visible only to the access groups chosen on the source — Snowflake's own role grants aren't read row by row; they only cap what the token, and so the connector, can reach at all.
Tokens expire on whatever schedule your admin set in Snowflake — the source stops syncing when one does, and pasting a fresh token over it resumes on the next sync. Records are capped the same way as the SQL source: the newest 5,000 rows per table.
Last updated 20 Sep 2026