TestGPT will (hopefully) find errors in your code

Startup CodiumAI releases TestGPT in beta. This AI system should help developers test their code.

New startup CodiumAI enters the arena with TestGPT. This is a new solution that, as the name suggests, is based on the GPT model of generative AI. More specifically, it is about OpenAI’s new GPT-4 model. TestGPT is not a debugger, but a true test assistant that can create interactive code tests. These tests not only test whether the application is running, but also whether it exhibits the desired behavior.

Taking the relevant tests is a time consuming and dependent process. The app’s stability is due in part to a meaningful and qualitative testing process. TestGPT should come up with related tests based on the written code. Then it’s up to the developer to accept the tests, and then the code goes through them.

Integration into development environments

TestGPT should complement other AI tools, such as GitHub Copilot. The new solution is currently available as a beta version. In time, the goal is to implement the AI ​​tool directly into development environments.

Generative AI models have taken off since ChatGPT was introduced to the general public. Projects like TestGPT show that the capabilities of such forms are not limited to text generation.