You still work serially. It’s just while you’re reviewing output or working on the next prompt you have other agent sessions doing work. It’s entirely possible to be in a flow state cycling between concurrent agent sessions constantly reviewing output, building context and honing prompts.
The context switching may be difficult for some. Those that can manage it and excel at these new skills will undoubtedly become more productive than those who cannot and do not.
Companies are already starting to require the use of these tools and cycling workflows. It won't remain optional for long, just as it's no longer optional per dev preference to use pull requests, to minimize unblocked work-in-progress, to achieve test coverage standards, etc.
Expectations of full agentic parallelization is still not so widespread though as employers are slow to pick up on the need to budget $2400-4800+ annually per developer for enough capacity with recent sota tools.