Morel is an ML dialect that can compile set-producing expressions into bytecode that Calcite can execute against databases

Sort of like "If you could query anything with SQL but it's ML instead"

I bring this up because the example query looks very similar to Morel queries

Neat xample of solving a combinatorial optimization problem with a single query that he posted recently to Twitter:

https://x.com/i/status/2062066151841321370

http://blog.hydromatic.net/2026/06/02/package-queries.html

I wonder though, is such a dialect better for agents? Have you tried to measure if an agent performs better expressing queries in such a language instead of SQL?

Not really, too many obscure symbols.

Certainly learnable but I wouldn't say immediately readable.