Both Akka Streams and Flow promote backpressured async stream processing.
But Flow’s purpose is to behave and provide a DSL to support Map/Reduce and Spark like jobs.
Akka Streams supports more foundational lower level concepts. There are standard components, Source, Sink, Flow, Fan-in, Fan-out, along with a graph based DSL that lets you compose these components together in arbitrary ways, while supporting standard functional transforms like, map, filter, etc over streams.
With a standard component model you can start to build an ecosystem of data processors and connectors. https://developer.lightbend.com/docs/alpakka/current/
If you like GenStage as a component model. Then these would be two DSLs that could sit on top of GenStage, providing a better developer abstraction.
You wouldn’t build the next generation of Phoenix on top of Flow. But you could/should with this new thing.