Highlights:

  • The business claims that its flagship product, Diffblue Cover, enables users to generate unit tests for Java code automatically every two seconds, which is 250 times quicker than a human developer.
  • According to Diffblue, the machine learning algorithm does not use the code for training, and it stays fully on-premises for security and privacy reasons.

An autonomous generative AI code test platform developer, Diffblue Ltd., raised USD 6.3 million in the latest funding round to create on its exclusive approach of generating unit tests for Java.

Along with previous investors – IP Group, Parkwalk Advisors, and AlbionVC, the new funding is an extension of the company’s Series A funding round. Oxford University also participated, and private investors contributed more capital through the Oxford Technology and Innovations EIS Fund.

Diffblue’s client Citigroup Inc. became a new equity investor in the round through its venture capital division, Citi Institutional Strategic Investments. The company has now raised over USD 46 million in total due to this financing.

Diffblue, which was founded by Oxford researchers, sets itself apart from existing generative AI systems by automating the error-prone portions of test creation in the software development lifecycle through reinforcement learning. It creates unit tests for Java code to ensure that it will compile and stay error-free. The business claims that it is error-free and free of hallucinations since the code it generates is deterministic and uses machine learning models rather than a large language model.

The business claims that its flagship product, Diffblue Cover, enables users to generate unit tests for Java code automatically every two seconds, which is 250 times quicker than a human developer. To produce software more quickly, developers can now devote their time to less time-consuming aspects of the process.

To ensure that each component of an application functions as intended, unit tests are a crucial software development technique. In essence, these tests are comparable to having a handbook that describes how to test every component of a car engine prototype to ensure that the distributor, fans, belts, spark plugs, and electrical system all function as intended before the vehicle is put into production. The tests make it easier to identify and address defects early in the development process.

Additionally, the AI makes code suggestions and automates improvements aimed at making Java code easier to comprehend and test. In this manner, it will be easier for human engineers to maintain, and subsequent tests will run more quickly. Because Cover is independent, it operates alongside developers, automatically creating updates and new tests whenever they make changes to their code.

“Most of the talk about generative AI in software development is really around a handful of massively capitalized companies pursuing LLM-based technology approaches to solve a broad range of use cases,” Diffblue Chief Executive Toffer Winslow stated. “Diffblue has chosen a different technical and financial path to address a more focused development problem, albeit one that is the bane of most software teams. The result of our focus is that we are able to solve this problem with a high degree of speed and accuracy without the security and privacy concerns that many companies often have with LLMs.”

According to Diffblue, the machine learning algorithm does not use the code for training, and it stays fully on-premises for security and privacy reasons. Since the working environment is completely local and protected by a firewall, no sensitive information may be leaked.

“Diffblue is leading the charge in one of the hottest, fastest-moving markets in the history of the software industry,” said Mark Reilly, Managing Partner at IP Group. “The company’s traction with large, sophisticated customers deploying best-in-class AI technologies across their software development processes is strong evidence of its differentiated, valuable solution.”

Key clients of Diffblue include ING Bank N.V., Cisco Systems Inc., AstraZeneca PLC, and the Bank of New York Mellon Corp. In addition to Citi, the company serves ten of the biggest U.S. banks and multiple Forbes Global 2000 companies.

According to the company, a new version of Cover will be released next month, making it more user-friendly for developers at all sizes of businesses.