To fully leverage development automation, tasks should already be defined on a technical level, ideally small enough for AI to implement and test in one go. This means POs will need new tools and workflows to produce better, more complete specifications. AI can help with this. We’ve used AI assistants that already know the project architecture, tech stack, user flows, and past tasks. They can help POs write clearer specifications and even ask the right questions to fill in missing details.
Designers, meanwhile, might skip the static mockup phase altogether. It makes little sense to design interfaces that developers later have to rebuild. With current AI tools, designers can already generate actual UI components with dummy data for demos. Developers can then focus on connecting real data and backend logic instead of wrestling with CSS alignment mysteries.
Developers, too, are crossing boundaries, especially into testing. Test-driven development (TDD) becomes even more powerful in AI-assisted workflows. When the implementation starts from tests, a large chunk of validation is already covered. AI can then analyze what’s been affected by changes and help decide what should be manually or automatically tested to ensure everything still behaves as expected.
These are just a few examples of how roles are shifting. As AI tools evolve, they allow people to step outside traditional boundaries, sometimes out of curiosity, sometimes out of necessity. And as AI boosts productivity, teams might shrink a bit, meaning each person takes on a wider range of work again.
