Meet Early: The AI That Catches Bugs Before They Bite
Early, a Tel Aviv, Israel-based startup that provides a generative AI code quality platform, officially launched earlier this month.
The company’s AI-powered solution, EarlyAI, acts as a test AI agent that automatically generates high-quality tests to help developers detect and fix bugs early in the development cycle. Since its soft launch in August, the platform has already generated over 30,000 unit tests and has been used by over 3,000 developers to improve the quality of their code, the company said.
“Early has enabled me to reach higher code coverage in my project with minimal effort,” Eddie Jaoude, an open source full stack developer focusing on JavaScript, TypeScript and DevOps, and founder of EddieHub, told The New Stack. “I even started to enjoy TDD [test driven development] as it allows me to write better quality code as well as giving me more time to do other aspects of coding which I enjoy.”
Early helps developers detect bugs at the beginning of the software development cycle by using GenAI to automate and enhance code testing, said Sharon Barr, CEO of the company. Early’s technology analyzes code, then automatically generates comprehensive unit test suites and validates them to ensure they effectively protect code and catch bugs.
“Some of the recent testing AI products out there are framed for testing needs only, whereas EarlyAI is looking at the broader software quality map, which I could see from the start,” Liran Tal, director of developer advocacy at Snyk, told The New Stack. “Specifically, you notice this as a user because the extension, prior to generating the tests for you, provides some callouts for adding DocBlocks style documentation where it is missing.”
The EarlyAI platform works within popular development environments like Visual Studio Code. Developers generate tests with a single click, review the generated tests, fix bugs, and see gaps in test coverage, Barr said.
“Something I at first didn’t expect and turned out surprisingly good was the website ‘add to vscode’ call-to-action button, which opens up my local VS Code instance and the EarlyAI extension on the marketplace,” Tal said. “I thought this would be some blog post landing page but it was a very smooth experience of taking me straight into the IDE to get value.”
Virtual AI Test Engineer
Early validates its technology through mutation testing, a process that assesses the quality of the generated tests by introducing small changes (“mutations”) to the code and verifying that the tests catch these changes, the company said. This ensures that the generated tests are comprehensive and effective at detecting potential issues.
“Unlike GitHub Copilot, we’re not just assisting developers in writing test code; we’re taking on the entire task of test generation, acting as an AI test engineer right alongside every developer,” Barr said in a statement. “Our goal is simple: we take care of the tests so developers can focus on what they do best — developing innovative applications. By catching bugs at the earliest stages of development, we aim to save companies billions in losses and accelerate the entire software development process.”
Time Saver
Richard Zampieri, principal engineer at Imdex Limited and creator of ExpressoTS, wrote in an article, “The first thing that struck me about Early was its ability to automatically generate unit tests for my existing codebase. Instead of crafting tests from scratch, I could focus on refining the generated tests and improving my code’s robustness and testability.
“This shift significantly accelerated my development process. The other interesting aspect I noticed is that 83% of the code generated I didn’t do any adjustment; it worked out of the box and increased my code coverage. Save me a huge time.”
Moreover, Zampieri wrote that in just 8.5 hours, he managed to:
- Generate unit tests for approximately 3,000 lines of code.
- Fix issues and enhance code testability.
- Achieve a total code coverage of 88% with 96 tests.
Origins and Plans for the Future
Early was founded by two former engineering leaders from EY. Barr’s co-founder, Lior Froimavich, most recently served as engineering lead at EY. The company currently has thousands of installations and hundreds of recurring users, Barr told The New Stack.
Lior Froimavich (left) and Sharon Barr of Early
The Early platform currently supports JavaScript andf TypeScript, with Python support in development. It uses OpenAI’s LLM technology but with additional proprietary layers to identify bugs in code through test generation.
“Using the LLM enables us to generate code, high quality code, like never before,” Barr told The New Stack.
He said Early provides 80% to 200% better results compared to existing solutions. Early positions itself as an “AI agent” rather than just a code assistance tool. It focuses specifically on test generation rather than general-purpose code and it includes deeper language understanding and context analysis.
Early is currently free for open source projects and the company plans to implement a freemium model in the future. The company is focusing on product-market fit before monetization, Barr said.
Early has secured some investment but is staying lean intentionally, Barr said. Earlier this month, the company raised $5 million in seed funding led by Zeev Ventures with the participation of Dynamic Loop Capital.
“Generative AI is the combustion engine of our era — a fundamental technology that will revolutionize every industry, including software development,” said Oren Zeev, founder of Zeev Ventures, in a statement. “Early is perfectly positioned to harness this power, focusing it on the critical challenge of code quality…”
Moreover, “they’re not just using AI; they’re applying it strategically to solve one of the most persistent and costly challenges in coding,” he noted. “I believe Early will become an indispensable tool for developers, dramatically reducing the time and cost associated with bug detection and fixing and ultimately accelerating innovation across the tech industry.”
Early’s future plans include expanding beyond unit tests to other types of testing, adding more programming language support, IDEs and types of testing, and building awareness and adoption over the next two to three months.
“We plan to take on the whole job of creating tests, acting as the AI test engineer working alongside every developer,” the Early team wrote in a blog post. “This will not only help developers avoid bugs but also free up their time and mental energy to focus on the creative aspects of coding that truly drive business value. It is the first step in our vision to reshape the software development landscape.”
