– Developers who build the code should not be the ones to test it
– Code testing has become a focus for generative AI startups
– Nova AI is breaking Silicon Valley rules by targeting mid-size to large enterprises with complex code-bases and not using OpenAI for their models
Code testing is an essential part of software development, but developers often dislike this task. As a result, there has been a rise in generative AI startups focused on code testing, such as Antithesis, CodiumAI, QA Wolf, and Nova AI. Nova AI, a year-old startup, has raised $1 million in pre-seed funding and is targeting mid-size to large enterprises with complex code-bases. The company uses AI technology to automatically build tests for continuous integration and delivery environments.
Nova AI founders, Zach Smith and Jeffrey Shih, drew their inspiration from their experiences as engineers at big tech companies like Google and Meta. They are breaking Silicon Valley startup rules by not relying heavily on OpenAI’s GPT models due to privacy concerns of large enterprises. Instead, Nova AI uses open-source AI models like Llama and StarCoder, as well as proprietary models, for code testing tasks.
The use of open-source AI models not only addresses privacy concerns but also proves to be cost-effective and efficient for specific tasks like writing tests. Nova AI believes that open-source models, like the latest version of Llama by Meta, can outperform GPT-4 and other domain providers when it comes to narrow, targeted tasks. The company’s focus on building and utilizing open-source AI models showcases the potential of alternatives to industry-leading models like GPT.