Generative artificial intelligence code quality startup Early Technologies Ltd. announced the availability of its VSCode extension today after closing on $5 million in seed funding.
Today’s round was led by Zeev Ventures and saw participation from Dynamic Loop Capital. The funds will help to accelerate the development of Early’s AI-powered Test AI agent, which is designed to automatically generate high quality tests of code, so developers can easily detect and fix any bugs in their software.
Using the company’s Test AI agent, developers can detect bugs at the very start of the software development cycle, the startup explains. It leverages generative AI to quickly analyze each new line of code written by the developer, in order to generate an appropriate unit test, which is then applied to the code to see if any bugs or vulnerabilities are present.
Early says its extension integrates with popular development tools such as Visual Studio Code and supports programming languages such as JavaScript, TypeScript and Jest. Developers can generate tests with a single click without leaving their coding environment, review the generated tests, see the results, fix any bugs it discovers and even see gaps in the test’s coverage.
To ensure the accuracy of its tests, Early uses a process known as “mutation testing,” which involves making small, intentional changes, or mutations, in the code to create a bug, and then verify that the test is able to catch it. If it does, then the test should theoretically also be able to catch unintentional issues within the developer’s code.
Early says it generates two kinds of tests. Green tests are designed to enhance coverage, while red tests are for identifying potential bugs within the code.
The company believes it can save organizations a lot of money by spotting vulnerabilities early through its automated tests. It cites a 2022 report by the Consortium for Information & Software Quality, which estimated that the cost of poor-quality software in the U.S. alone had grown to more than $2.41 trillion, thanks to the downtime and cyberattacks that result from it.
Early stresses the importance of being able to catch bugs as soon as possible. It points out that the impact of deficiencies in software escalates dramatically when they are discovered much later on in the development cycle, especially if they’re only found after the software has been released. A case in point is the recent global information technology outage caused by CrowdStrike Holdings Inc., when it rolled out a faulty update to its popular Falcon cybersecurity software, causing millions of Windows-based systems to crash, disrupting the operations of hospitals, government agencies, airlines and numerous businesses worldwide.
According to Early, the CrowdStrike debacle was caused by a relatively simple “out-of-bounds exception” that could have been easily detected by one of its tests. Yet it slipped through the net, resulting in billions of dollars in losses. The incident highlights the critical need for improved code quality, the startup said.
Early’s co-founder and Chief Executive Sharon Barr (pictured, left, alongside co-founder and Chief Technology Officer Lior Froimavich) said a lot of effort has been made in using generative AI technology to develop software code, but much less focus on the task of test generation for that code.
“We’re acting as an AI test engineer that works right alongside every developer,” he said. “Our goal is simple: We take care of the tests so developers can focus on what they do best. By catching bugs at the earliest stages of development, we aim to save companies billions in losses.”
More than 3,000 developers have already began using Early’s VSCode extension since its soft launch in August, and it has generated more than 30,000 unit tests already, the company said.
Zeev Ventures founder Oren Zeev said he’s backing Early because it’s taking on a critical challenge in trying to improve code quality. “Early is not just using AI, but applying it strategically to solve one of the most persistent and costly challenges in coding,” he said. “I believe it will become an indispensable tool for developers.”