LambdaTest launches KaneAI agent for end-to-end software testing

California-based LambdaTest, the company known for helping leading enterprises test how their apps work across different platforms, is expanding into the AI domain with the launch of KaneAI, an agentic experience for end-to-end software testing and quality assurance.

Available to select partners as an extension of the core LambdaTest platform, KaneAI allows users to write, execute, debug and evolve automated tests using nothing but natural language.

It comes as a major upgrade from the six-year-old company, paving the way for users to get past complex coding and low code workflows when authoring and managing tests.

“The (LambdaTest) engineering and product teams…pushed the boundaries of what is possible in quality engineering. They ensured that every aspect of KaneAI could meet the real-world challenges that testing teams face daily. This journey was about developing an AI-powered test agent and reimagining what test automation could be.,” Asad Khan, the co-founder and CEO of the company, said in a statement.

What to expect from KaneAI?

Since its inception in 2017, LambdaTest has made a name for itself by providing quality assurance teams with a unified cloud for cross-browser testing. This allows users to set up automated or manual tests to check how their applications work across thousands of different browser and OS configurations, both on desktop and mobile. 

“The cloud-based platform enables users to run tests up to 70% faster than any other infrastructure and allows users access to a diverse set of testing environments without needing to invest in physical hardware. This not only significantly reduces costs and increases efficiency, but also provides the scalability and performance required to run large-scale quality assurance workflows,” Khan told VentureBeat.

As a result, teams can ship out high-quality, bug-free releases to keep up with the fast-moving delivery cycle.

However, as the adoption increased, LambdaTest realized that providing a unified platform for test automation across different environments might not be enough.

They also need to make it approachable and easier to use, going beyond complex approaches of writing custom test scripts or integrating low-code solutions that have a steep learning curve and begin to break down at scale. 

This led to the work on KaneAI, which leverages generative AI to enable end-to-end test automation, covering all the steps from writing and executing tests to reporting and debugging them with minimal effort.

With the web AI agent, QA teams can write test steps for a particular comprehensive action in natural language (like searching for hotels in a particular city during a given timeframe). When the steps are run, the agent’s underlying models analyze and execute them – one at a time – in a cloud browser visible to the user.

If the task is too complex and the user is struggling to express it in words, they can even use an interactive mode to take the action in the browser window, allowing the AI agent to record and convert it into a text step. 

Once the steps are executed, the entire test case can be added to LambdaTest’s test manager. From there, the user can generate the associated test code in their preferred language and framework and run it on the company’s HyperExecute cloud, which intelligently groups and distributes tests across different environments, orchestrating test execution based on past run data. KaneAI also ties to LambdaTest’s tools that provide users with granular insights into the automated tests, covering reports, metadata and debugging features. 

Under the hood, KaneAI uses OpenAI’s models and those internally trained by LambdaTest to deliver the agentic testing experience.  

“We are leveraging data from billions of tests executed on our platform to develop the platform with seamless end-to-end test creation and features like dynamic 2-way test creation with code-to-instruction and instruction-to-code translation,” Khan added. 

While there are plenty of generative AI-powered coding tools and agents – including Cognition’s Devin – that can run tasks based on commands, Khan says KaneAI differentiates by going deeper and enabling users to manage the entire testing journey on a single platform. 

“One of the reasons for building KaneAI has been that current tooling in the market is not on par to provide a holistic testing experience. Even with the current code generation capabilities of AI, incorporating end-to-end capabilities for a continuous testing process like CI/CD integration, report generation, test analysis, and debugging involves juggling multiple tools which adds layers of complexity,” he noted.

That said, the CEO also pointed out that the web agent is not yet available for enterprises within the LambdaTest ecosystem. The company is currently beta-testing it with select customers, power users and industry experts and has plans to expand access to waitlisted users over the coming months – with necessary improvements. 

“There are a lot of use cases that can be done using the platform but the full capabilities are yet to be discovered. For example, we are adding more integrations to the platform right now that will enable users to run tests from platforms like Slack and Microsoft Teams,” Khan said while adding that the natural language-based offering will also make it easier for business stakeholders to become an integral part of the testing process.

Currently, more than 10,000 organizations, including the likes of Nvidia, Vimeo, Microsoft and Rubrik, use LambdaTest to run millions of daily tests. The number is substantial, but it’s not an easy ride for the company as the space has multiple other players, including heavily funded BrowserStack and Sauce Labs as well as open-source Testsigma.