Glossary · AI & ML

Join Inference

Join inference is the ability of an AI query system to automatically determine which tables to join, and on which keys, to answer a question that spans multiple tables.

Real questions almost never live in one table. "Revenue by customer segment" touches orders, order items, customers, and a segments table–and connecting them requires knowing the join keys and the right join types. Join inference automates this: the system reads foreign key relationships from the schema (or infers them from naming patterns when constraints are missing) and constructs the join path a careful analyst would write. This matters because joins are where non-experts give up on SQL and where subtle bugs hide–an accidental row-multiplying join can double-count revenue without throwing any error.

Fig — every answer ships with the tables, rows and SQL behind it.