The Snowflake integration replaces SFTP/CSV-based data delivery with a direct connection to your Snowflake warehouse. Cora.ai connects using RSA key-pair authentication, queries your Snowflake views on a schedule, ingests the data, and makes it available to agents, dashboards, and workflows.

Integration Type: Key-Pair Auth (RSA Private Key), Organization-Level
​Required Role: Snowflake Account Administrator or equivalent
​Why Key-Pair?: RSA key-pair is the industry-standard authentication method for Snowflake service accounts — recommended by Snowflake and used by dbt, Fivetran, Airbyte, and other major data tools — because it never expires, requires no token refresh, and transmits no secrets over the wire.

Note: The steps below represent our recommended setup for security and isolation. Your team is free to adapt the naming, schema structure, and warehouse configuration to fit your existing conventions — the only hard requirements are that Cora.ai's service user has SELECT access to the relevant views and uses key-pair authentication.

We recommend a dedicated database and schema that contains only the views Cora.ai should access. This acts as a clear security boundary — Cora.ai has no visibility into any other schema. If your team prefers to grant access to views in an existing schema, that works too.

CREATE DATABASE CORA_INTEGRATION_DB;

CREATE SCHEMA CORA_INTEGRATION_DB.CORA_VIEWS;

Then create views in this schema that mirror your data exports:

CREATE VIEW CORA_INTEGRATION_DB.CORA_VIEWS.ACCOUNTS AS

SELECT account_id, name, industry, created_at

FROM PRODUCTION_DB.CORE.ACCOUNTS;

Keep the same column names and data types so the schema Cora.ai sees is consistent with your source data.

Create a dedicated role and service user scoped exclusively to the Cora views schema.

-- Create a role with no inherited privileges

CREATE ROLE CORA_INTEGRATION_ROLE;

-- Create a service user (cannot log in interactively)

CREATE USER CORA_SVC

DEFAULT_ROLE = CORA_INTEGRATION_ROLE

DEFAULT_WAREHOUSE = CORA_WH

TYPE = SERVICE;

-- Grant the minimum required privileges

GRANT USAGE ON DATABASE CORA_INTEGRATION_DB TO ROLE CORA_INTEGRATION_ROLE;

GRANT USAGE ON SCHEMA CORA_INTEGRATION_DB.CORA_VIEWS TO ROLE CORA_INTEGRATION_ROLE;

GRANT SELECT ON ALL VIEWS IN SCHEMA CORA_INTEGRATION_DB.CORA_VIEWS TO ROLE CORA_INTEGRATION_ROLE;

GRANT SELECT ON FUTURE VIEWS IN SCHEMA CORA_INTEGRATION_DB.CORA_VIEWS TO ROLE CORA_INTEGRATION_ROLE;

-- Dedicated X-Small warehouse with auto-suspend (cost isolation)

CREATE WAREHOUSE IF NOT EXISTS CORA_WH

WAREHOUSE_SIZE = 'XSMALL'

AUTO_SUSPEND = 60

AUTO_RESUME = TRUE

INITIALLY_SUSPENDED = TRUE;

GRANT USAGE ON WAREHOUSE CORA_WH TO ROLE CORA_INTEGRATION_ROLE;

-- Bind role to user

GRANT ROLE CORA_INTEGRATION_ROLE TO USER CORA_SVC;

Generate an unencrypted RSA private key and assign the public key to the service user:

# Generate private key (unencrypted PKCS#8 PEM)

openssl genrsa 2048 | openssl pkcs8 -topk8 -inform PEM -out cora_svc_key.p8 -nocrypt

# Extract public key

openssl rsa -in cora_svc_key.p8 -pubout -out cora_svc_key.pub

Assign the public key in Snowflake:

ALTER USER CORA_SVC SET RSA_PUBLIC_KEY='<paste public key without BEGIN/END lines>';

See Snowflake's Key-Pair Authentication guide for details.

Once the Snowflake integration card shows Connected, Cora.ai will begin querying your views on a scheduled basis. Confirm the expected data is flowing correctly in Cora.ai.

Enable data ingestion to sync information from this integration into Cora.