Technology

FoxPhone: Isolated cloud Android devices for AI agents and QA fleets

FoxPhone provides isolated cloud Android devices for AI agents and QA fleets. Each session starts on its own device, ensuring accounts, cookies, and fingerprints do not mix.

FoxPhone: Isolated cloud Android devices for AI agents and QA fleets

FoxPhone provides isolated cloud Android devices that behave like real smartphones but run entirely in the cloud. Teams running AI agents, quality assurance, and app testing can launch parallel Android environments without purchasing, storing, or resetting physical devices after every run. The platform is aimed at developers, QA leads, and operators of agent fleets who require reproducible sessions while preventing accounts, cookies, and device fingerprints from mixing between tasks.

Many test labs still operate using shared phones. One agent logs in, a second takes over the same device, and remnants of the previous session remain in memory, the browser profile, or app data. This leads to flaky tests, blocked accounts, and results that cannot be replicated the following day. FoxPhone isolates every session onto its own cloud device. The task starts on a fresh Android profile, works only with the data intended for that run, and can be discarded or archived afterwards. This keeps test history traceable without traces bleeding into the next run.

For AI agents, this separation is particularly important. An agent operating an app, filling out forms, or verifying a checkout creates side effects: logins, push tokens, cache files, and intermediate states. If multiple agents share the same device, they influence one another. Isolated cloud Android devices give each agent their own interface, their own identity, and their own lifecycle. Operators can spin up fleets in parallel, distribute tasks, and decide at the end of each session whether the device should continue running, pause, or be deleted. This reduces wait times in the pipeline and makes load spikes manageable, as a physical device wall is no longer the bottleneck.

Quality assurance benefits from this same architecture. Manual testers and automated scripts can request the same device types, lock in screen resolutions, and document bugs with a clear device context. Instead of grabbing a phone from a cupboard and hoping no one else has been working on it, the team books a cloud device for that exact case. Regressions can be repeated on a clean slate. Exploratory tests can run on a copy without altering the reference environment. If a build fails, the state is preserved until the team has secured the screenshot, logs, and sequence of steps.

FoxPhone is accessible via the browser. There is no need for additional hardware rollouts, and new team members do not need test phones shipped to them. Those who wish to see how it works can find the product description on the FoxPhone website. For teams specifically orchestrating devices for tests, the Cloud Device Testing section describes how sessions are set up for QA and agents. Both pages explain the use of isolated Android instances without a shared device pool mixing accounts or fingerprints.

A typical workflow looks like this: First, the team defines the device type and the purpose of the session, such as a login scenario, a checkout, or an agent task with multiple app switches. Then, FoxPhone starts an isolated cloud Android device. The tester or agent connects, performs the steps, and saves the evidence. Finally, the session is terminated. The next task receives a clean device, even if it is testing the same account type or app. This repeatability is the core of the offering: identical start conditions, separate data, traceable results.

Companies bringing agents into production also face questions of governance. Who used which device and when? What data was visible during the session? Could one task see the identity of another? A fleet of isolated devices answers these questions more clearly than a shared phone. Each session has a boundary. Access credentials do not need to remain on a shared device. If an experiment goes wrong, it only affects that single device and not the entire lab fleet.

FoxPhone LLC provides the platform for teams that want to operate Android in the cloud without compromising on session isolation. Michael Johnson describes the focus as follows: fewer shared hardware, more controlled cloud devices, and a workflow that is sustainable for both human QA and AI agents. Interested parties can learn more about the offering on the FoxPhone homepage and read about how cloud devices are used for tests and agent fleets on the Cloud Device Testing page.

The press release is aimed at product teams, test engineers, and operators of automated Android workflows. It does not replace individual consultation and describes the current product approach: isolated cloud Android devices, parallel fleets, and clean sessions for AI agents and QA. Further details are available on the public website. Teams are not just comparing the number of devices, but the quality of the isolation: separate profiles, separate session data, and a clear conclusion after every task. These specific features make cloud Android suitable for daily use in agent fleets and QA labs when physical phones no longer scale.