Your database is full of answers.

Baseil finds them. It maps your tables on its own and answers in plain English, with the query that ran.

Each dot is one row of a sample store database: 1,817 rows in 14 tables.

It sorts itself out.

Baseil reads every table, column, and relationship, then builds read-only query tools on top. Nobody writes a schema doc.

0 tables, 0 columns, 0 relationships

Then you just ask.

Which product categories get returned the most?

Apparel gets returned the most.

0 of 108 order lines came back. Every dot that moved is a real row: each return found its order line, each line found its category, and the stacks are the count.

Joined categories › products › order_lines › returns

The query that ran

SELECT c.name AS category,       COUNT(ol.id) AS lines,       COUNT(r.id) AS returns,       ROUND(100.0 * COUNT(r.id) / COUNT(ol.id), 1) AS return_rateFROM categories cJOIN products p ON p.category_id = c.idJOIN order_lines ol ON ol.product_id = p.idLEFT JOIN returns r ON r.order_line_id = ol.idGROUP BY c.nameORDER BY returns DESC;

Get all your data talking.

Ask in chat, or let your agents ask over MCP. Every answer comes with the query that ran.

Install

curl -fsSL https://releases.baseil.ai/install.sh | sh