Requirements

Private Preview

Summary of everything your team needs to provide or setup to enable the Ekai native app.

Setup & Identifiers

Consumers are requested to do the following in order to enable and access Ekai:

Share your Snowflake account details: region, locator, and account identifier
Provision access to LLMs by ensuring at least one of the following:
1. Anthropic, OpenAI, and Google standard API keys to connect to hosted models.
2. Use secure, private endpoints via Google Vertex AI or Azure OpenAI.
- Azure OpenAI — gpt-5-mini, text-embedding-3-small
- Google Vertex — gemini-3.1-pro-preview, claude-sonnet-4-6, claude-haiku-4-6
Describe the target data (tables, row/column counts, total size) so Ekai can scope an efficient end-to-end runtime for the trial.
From the Snowflake UI, click GET to provision the app in your account and region.

ACCOUNTADMIN role is required to install the app.
Account Admin must grant application roles to end users.
The app creates an Ekai DB — ensure users have access. By default, all users with assigned application roles have full database access.
The app must be granted access to target objects: databases, schemas, tables, views, and warehouses. See Data Access Configuration.
An external user account should be created and shared with the Ekai team for setup, testing, and debugging.
Optional: Ekai allows users to connect to their configured data via the app directly, or by generating Programmatic Access Tokens (PAT) for user specific access.

Verify the following before starting the trial:

Object	Recommendation
Warehouses	Large for tens of GBs / billions of rows; XL for larger datasets
Snowflake Intelligence	Ensure SF Intelligence is enabled and required LLMs are accessible. Alternatively, create the Intelligence object with cross-region inference.
Event Tables	An event table must exist for log access (usually present by default — account admins can verify).
Cortex Analyst Ownership	If not using a PAT, users may need to enable usage of Snowflake Intelligence objects.