We gave cloud agents a phone to QA

A reliability engineering team has equipped cloud coding agents with iOS simulators and Android emulators so they can test mobile pull requests, capture screenshots and recordings, and attach evidence-based QA reports for human reviewers. The system runs each agent in an isolated environment with its own backend, database, seed data, and optional dependent backend changes, allowing end-to-end testing without affecting shared systems. To make the app automatable, the team improved accessibility roles, labels, test identifiers, deep links, and development authentication, enabling agents to locate controls and reach relevant screens quickly. Reports link each claimed UI behavior to screenshots and video timestamps, disclose untested areas, device and build information, model usage, timing, and cost, while leaving approval decisions to people. In its first two weeks, the system QA’d 42 pull requests, with most runs initiated by the agents that wrote the changes; typical runs took about 11 minutes, and QA-only runs had a median cost of roughly $7. Future work includes making QA automatic for all mobile changes, flagging stale evidence after new commits, expanding seed data, isolating QA costs, and supporting on-demand native builds.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.