Snowflake Integration

What it does

Query your data warehouse for usage events, analytics tables, and product telemetry.

How your data reaches Specky

Specky pulls from Snowflake on a schedule. A sync runs when you open your workspace if the data has gone stale, and you can trigger one by hand from the integration card at any time.

What you'll need

Account Identifier (Required) — Found in Snowflake → Admin → Accounts.

Warehouse (Required) — The virtual warehouse to run queries against.

Database (Required) — Default database to query.

Private Key / Password (Required) — Use key-pair auth (JWT) or a service account password.

Setup

Go to Settings → Integrations, find Snowflake, and click Connect.

Create a service account user with USAGE on the warehouse and SELECT on target schemas

Generate a key-pair and copy the private key

Fill in the account identifier (format: orgname-accountname or locator.region)

Data synced

Table metadata from your INFORMATION_SCHEMA

Configure custom SQL queries via the AI chat after connecting

After connecting

Everything Specky pulls in from Snowflake shows up in Discovery alongside your other sources, and feeds synthesis, clustering, and the AI chat's answers about your product. You don't have to do anything with it directly — it's there so the rest of Specky has something to reason over.

Disconnecting Snowflake removes both the connection and the signals it brought in.

Troubleshooting

Connected, but nothing appears. Open the Snowflake card and run a sync by hand. If it errors, the card shows the reason.

"Upgrade to connect". Your plan caps how many integrations you can connect at once, and a few providers are Scale-only. Settings → Usage shows where you stand.

Credentials stopped working. Tokens expire and get revoked. Reconnect with a fresh one — your existing signals are kept.

See the Snowflake integration page for what it's for and the questions people usually ask first, or the Integrations Overview for how integrations work in general.