Draxlr adds customer-facing analytics to a SaaS product without building a reporting layer in-house. It queries your SQL database directly, with no pipeline, warehouse or data copies.
Your backend requests a short-lived embed token using your API key and secret, so the secret never reaches the browser. Row-level security attributes such as a tenant ID are signed into that token and applied to every query. If an attribute is missing, the query returns no rows. Embeds ship as HTML snippets, React and Vue SDKs, or backend APIs, and dashboard edits appear without a redeploy.
The full app embed lets your customers write their own SQL or ask the AI assistant questions in plain English. Those queries are parsed, limited to read statements and restricted to the customer’s own rows.
An MCP server lets Claude, ChatGPT and Cursor query your databases read-only. Supported databases include PostgreSQL, MySQL, SQL Server, Snowflake, BigQuery, Databricks and ClickHouse.