TestGPT – AI Tool Validating Code Integrity.

Shri Krishna @SriTalkstech

Today, Tel Aviv-based Codium AI released a beta version of its generative AI-powered code-integrity solution, dubbed TestGPT. Designed to assist developers in testing their code, the TestGPT model offers autogenerated software test suite suggestions for developers to speed coding and bug scans, starting with Python and JavaScript. 

Codium helps developers automate the all-important test creation process. The company said it received $11 million in seed funding to develop this AI model. 

Addressing Software Failures

The potential for such a tool is significant. In 2020, the cost of software errors in the U.S. alone was a staggering $2 trillion, leaving many companies questioning the quality of their software. Errors propagate throughout the software development life cycle, and the cost of addressing them compounds. But software testing is a laborious and time-consuming process.

Having led product and R&D teams at companies like Alibaba Cloud, Itamar Friedman and Dedy Kredo understood these challenges firsthand. Backgrounds in software development, machine learning and product management convinced them of the potential of AI large language models (LLMs) for software test validation, and they built Codium AI in 2022. 

Writing non-trivial test cases is tedious and frustrating, he said. “Sometimes you even hate writing tests, but the alternative of letting a bug get into production can be a disaster.”

Codium’s first tool is an IDE (integrated development environment) extension that enables an iterative process of generating tests and then tweaking code based on the outcomes of those tests. This interaction with the developer helps the tool understand the code better and generate more accurate and meaningful tests, while guiding the developer to write better code. 

Like ChatGPT, Copilot and other generative dev tools, the TestGPT system exploits generative AI models. But TestGPT is focused on verifying the correctness of code versus the desired specification, according to Friedman. It is meant to enable high code integrity so developers can develop faster. “It embeds testing best practices in its prompting process, and does a series of pre- and post-processing steps to ensure high-quality outcomes,” Friedman said.

Leave a Comment

Your email address will not be published. Required fields are marked *