With Datasphere, AIs tend to use the full HANA functionality, but only a limited set of HANA functions is available in Datasphere.
With SAC it is even more difficult, as AIs tend to use a standard JavaScript approach, whereas SAC uses its own APIs and coding. But now we have built-in AI-assisted generation of SQL in Datasphere and JavaScript in SAC. The SAC feature has been available since June 2026, while the Datasphere feature came out just before TechEd.
So how does it work? Activate it for free and use it :). As with SAC’s assisted JavaScript generation, no AI Units are consumed.
In Datasphere, just drag and drop the objects you would like to use in the new SQL. Then write down what you need. The difference from J4C or Claude is that you no longer need to feed in the metadata (burning input tokens to provide the context), because Joule is already grounded.
Made a mistake while dragging and dropping? No problem. Remember, Joule is already grounded and automatically fetches the correct source object.
Don’t know which keys to use? No problem. Joule has access to the metadata and is a fully trained LLM for SQL, so just tell it to join whatever it sees fit.
Of course, as with any AI, a human needs to stay in the loop!
This feature makes ETL development fun. I can concentrate on delivering the solution instead of wasting time on missing commas, misspellings, incorrect joins, ...
Bluesky
