Software developers can now type a test in plain English and watch it play out on a real Android phone or iPhone sitting in a data centre. TestMu AI has switched on real-device support in Kane CLI, its artificial intelligence (AI) testing tool that runs from the terminal, the text window where developers type commands.
Until this release, Kane CLI ran mobile tests only on simulated phones, and running those locally needed a Mac with Apple’s own chips plus Apple’s Xcode or Google’s Android Studio. The company says real-device runs work from Mac, Windows and Linux machines, with nothing to set up beyond signing in to an eligible TestMu AI account.
“Real device testing has always been the part of mobile QA [quality assurance] that teams put off because it needed hardware, a Mac, and a separate tool,” said Mudit Singh, a co-founder of TestMu AI. “We wanted it to be one line in the terminal,” he added.
The company is based in San Francisco, with offices in Noida and Bengaluru in India and in Dubai. It traded as LambdaTest until January 2026, and this update ties its newest AI tool to the phone cloud it built under that name.
Picking A Phone And Watching It Run
A developer chooses Android or iOS, the operating system on iPhones and iPads, with a single setting. They can name a phone model and an operating system version, give just one of the two, or leave both out and let TestMu AI’s cloud assign any free device.
While the test runs, the phone’s screen appears in a separate, view-only window on the developer’s laptop and updates with each tap. Kane CLI accepts the standard Android build files and iPhone app packages, and the company says each build is uploaded once and then reused. It also says the tool deals with Android permission requests and iOS pop-up alerts by itself.
Every run is saved, so a passing test can be replayed later without the AI working out each tap again, and one command repeats it across many devices in the cloud. Steps and screenshots land in TestMu AI Test Manager, its results tracker. The documentation flags one gap: “adaptive heal”, which tries to repair a broken step during a replay, does not work on real devices.
Built For Terminals And Coding Agents
Cloud runs in Kane CLI are paid for through KaneAI, the company’s AI testing agent, which went on general sale in September 2025. Instead of writing code that hunts for buttons on a screen, a developer describes in everyday words what a user would do and what should happen, and the AI carries out the steps.
TestMu AI built the tool for AI coding agents, the assistants that write software for developers, as much as for people. Its product page lists Claude Code, Cursor, Codex and Gemini CLI among the agents it works with. An agent setting turns the output into a structured format that software can read, and simple exit codes tell an automated build system whether a test passed, failed, hit a setup or login error, or timed out.
Each run also leaves an evidence file holding the steps, screenshots, logs and final verdict. If an app shows the wrong total, for example, the company says Kane CLI flags a bug instead of bending the test until it passes. Finished web flows can be exported as Playwright code and mobile flows as Appium scripts in Python, two widely used open-source testing frameworks.
The design reflects how much code AI tools now write, all of which still has to be checked before release. Younger companies are chasing the same problem, among them Zurich’s Klarent, whose AI agents write app tests and keep them up to date.
Plans, Prices And The Device Cloud
Kane CLI costs nothing to install and use, but anything run on the company’s cloud is charged to a TestMu AI plan. Real phones are open to customers on KaneAI Max, which the documentation says includes one real device, or to those who pair KaneAI credits with the Real Device Plus Automation Cloud plan. Both routes need Kane CLI version 0.9.0 or later.
The company’s pricing page lists KaneAI Max at $179 a month and Real Device Plus Automation Cloud at $199 a month, both billed yearly. KaneAI’s entry plan, which runs tests only in the developer’s own browser, starts at $17 a month on the same yearly terms.
The phones themselves sit in what the company says is a cloud of more than 10,000 physical iOS and Android devices. TestMu AI lists Apple iPhones and iPads alongside Samsung, Google Pixel, OnePlus, Xiaomi, Oppo, Vivo and Motorola handsets, and its live-testing plan lets testers work on up to six phones side by side.
From LambdaTest To TestMu AI
The company made its name in cloud testing, letting teams check websites and apps across thousands of browser and device combinations, a market where it competes with BrowserStack and Sauce Labs. It now calls itself a full-stack agentic AI quality engineering platform, meaning AI agents take on much of the testing work. The new name borrows from TestMu, the free online testing conference it has hosted since 2022.
Asad Khan, co-founder and chief executive, runs the company with fellow co-founders Jay Singh, Mayank Bhola and Mudit Singh. TestMu AI says its platform has more than 3M users and over 18,000 enterprise customers in 132 countries, and it names Microsoft, OpenAI, Nvidia, GitHub, Workday and Akamai among them.
Its last funding came as LambdaTest, which raised $38M in a Series D round led by Avataar Ventures in December 2024, with Qualcomm Ventures taking part. That took its total to $108M, from backers that also include Premji Invest, Peak XV Partners and Blume Ventures.
Simulated Phones And Their Limits
Simulators and emulators copy a phone in software on a computer, which makes them quick and cheap for everyday checks. Google’s own Android guidance still tells developers to try an app on a real device before it reaches users.
TestMu AI made the same point in September, when one of its blog posts said emulators and simulators, including those Kane CLI used at the time, miss behaviour tied to specific hardware. That post sent teams to separate tools such as Appium, Espresso or XCUITest for tests on physical phones.
Kane CLI began by testing websites in a real Chrome browser, had added simulated phones by August and now reaches physical handsets in the company’s cloud. The features, device numbers and customer figures all come from TestMu AI. For developers on an eligible plan, a real phone is now a terminal command away rather than a drawer of handsets.
