Meet William: The Best AI Agent for Testing Games & Apps on Real Devices

Say Hello to William 👋

The best AI agent that tests games and apps on real devices.

William aims to solve one problem: the countless hours people spend every day tapping buttons on phones, playtesting mobile games, running research across apps, and doing other manual work nobody likes doing.

We think that’s a problem worth solving. We built William to take that work off your plate, in a much more natural way.

Okay so, what is William?

William is an AI agent built by nunu.ai (us!) that operates real devices (phones and PC) just like a human would. It looks at the screen, taps, swipes, and navigates apps to perform tasks like QA testing, data collection, and workflow automation.

Unlike traditional mobile test frameworks like Appium, Espresso, or XCUITest, which require engineers to write and maintain scripted test cases, William takes natural-language instructions and executes them autonomously on physical devices.

How does it work?

Step 1: Upload your Build. Upload your APK or IPA to nunu.ai and simply tell William what to test in plain language. No scripts, selectors, or setup required.

Step 2: Run your task. William then connects to a real phone and executes the task directly on the device, exactly the way a human would.

Step 3: Review the results. Each run produces a full recording showing every action William took together with it's reasoning and additional insights.

For example, William can run a menu test for a mobile game by connecting to the phone, finding and opening the app, navigating the menus, and surfacing any bugs along the way.

Beyond QA

William isn't limited to bug hunting. Teams are also using it for:

• Competitive research → collecting data from competing apps that don't expose APIs and summarize them in a structured report.
• Document and report generation → pulling structured outputs from in-app workflows (f.e. creating Design Documents).
• General phone automation → Let William perform custom workflows across phones and PCs.

Run Multiple AI Agents in Parallel

With nunu.ai, you can spin up as many Agents as you need, each connected to its own physical device, allow for multi-agent workflows like testing multiplayer games. Whether you're testing a single build or running thousands of parallel sessions across device types, the infrastructure scales with you.

That means faster QA cycles, broader device coverage, and the ability to catch issues that scripted tests routinely miss.

Try William: Now in Public Beta

William is live in public beta today. If you're building games, apps, or anything that runs on a phone, you can start using it now at nunu.ai.