ai_query
A built-in SQL function provided by Databricks that allows users to invoke deployed Model Serving endpoints directly within standard SQL statements. It encapsulates engineering details such as HTTP request construction, result parsing, and error retry logic.
Timeline (last 90 days)
ai_query is a built-in SQL function provided by Databricks that allows users to directly initiate inference requests to deployed Model Serving endpoints within standard SQL statements.
The calling syntax of ai_query resembles SELECT ai_query('endpoint_name', input_column) FROM my_table, with the function internally handling HTTP request construction, result parsing, and error retries.
Lakeflow is Databricks' data pipeline orchestration tool, which natively supports scheduling, monitoring, and dependency tracking of SQL tasks containing ai_query as a node within a DAG.
When performing inference within the same platform via ai_query, Unity Catalog's governance capabilities can fully cover model invocation behavior, making inference auditable and traceable.
All Facts (4)
ai_query is a built-in SQL function provided by Databricks that allows users to directly initiate inference requests to deployed Model Serving endpoints within standard SQL statements.
50%UnverifiedThe calling syntax of ai_query resembles SELECT ai_query('endpoint_name', input_column) FROM my_table, with the function internally handling HTTP request construction, result parsing, and error retries.
50%UnverifiedLakeflow is Databricks' data pipeline orchestration tool, which natively supports scheduling, monitoring, and dependency tracking of SQL tasks containing ai_query as a node within a DAG.
50%UnverifiedWhen performing inference within the same platform via ai_query, Unity Catalog's governance capabilities can fully cover model invocation behavior, making inference auditable and traceable.
50%