AI is changing how React Native apps get built, tested, and reviewed

Zach moved from mechanical and structural engineering into software after building small programs and a device to control his apartment’s air conditioner. Learning React helped him see how components and state could make building interfaces feel direct, and he later moved into React Native.

Zach uses agents differently depending on the task: sometimes he codes himself, and sometimes he asks an agent to explore an approach or gather documentation. He now checks agents’ claims carefully, since one incorrect detail among many correct ones could cause a production issue. He sees AI as a way to understand and improve code together, rather than simply offloading work.

Before device automation, an agent could suggest a change, but Zach had to test it, inspect logs, and report back before it could continue. With Agent Device, agents can interact with iOS and Android simulators, collect screenshots or recordings, and verify changes while Zach works on something else. He describes changing the iOS workflow so the simulator runs in the background after its first launch.

Zach describes using Cursor Cloud Agent from his phone and seeing a recording of its work on an Expo project. He also built a feature from a phone in a cloud session, then asked Codex to review it and used end-to-end tests in CI to check Linux support without a Linux machine. The hosts discuss running QA agents in CI to produce screenshots for reviewers and help teams cover more app changes.

Zach expects AI to assist across product planning, development, code review, and QA, while people and agents continue to work together. He says agents can generate more code than he would write, so he spends time asking them to simplify and clean it up. He closes by encouraging new developers to build and stay open-minded, while noting that deciding which problems are worth solving still matters.