Parallel execution as part of a language's basic execution model

Consider an application that needs three independent results: Data from an API. A database query. A computation based on an input file. A sequential implementation waits for each operation to finish before starting the next. A concurrent implementation needs to manage independent tasks and combine their results. OcaltQL has a parallel execution model in which AND branches can execute operations concurrently. When multiple branches assign the same variable, the variable becomes a multivariable. COLLAPSE can then collect its values into an ordered array. Documentation: ql.ocalt.com/tutorial/multivariables-parallel This model is interesting because parallel execution is expressed directly in the language rather than exclusively through external concurrency libraries. It does not eliminate the need to consider shared resources, race conditions, or operation dependencies. But it raises an interesting design question: What would application code look like if parallel execution were a first-class part of the language rather than an additional abstraction layered on top?