Diffblue, founded by researchers from the University of Oxford and based in Oxford, England, builds reinforcement learning–powered AI tools to automate unit test generation. Its flagship product, Diffblue Cover, autonomously generates unit tests that are guaranteed to compile and pass while delivering high coverage and test strength. The company combines reinforcement learning with code execution to refine outputs, reduce hallucinations, and improve accuracy versus traditional LLM approaches. Diffblue says Cover can write unit tests up to 250x faster than a human developer and is used by customers including Citi, ING, Workday, S&P Global, and BNY. Through participation in the ITEA GENIUS project, Diffblue plans to develop tools and processes to automate complex software-engineering tasks, improve software quality, and accelerate development cycles. The company recently received a £1 million grant from Innovate UK to accelerate enhancements to its reinforcement learning–powered solutions for autonomous test code generation. Diffblue builds Diffblue Cover, an autonomous generative AI-for-code product that automates unit test generation using reinforcement learning techniques rather than LLM-based coding assistants. The company says this differentiated approach helps avoid accuracy, security, privacy, and liability concerns associated with LLMs. Diffblue Cover aims to improve code quality, expand test coverage, and increase engineering efficiency so teams can ship software faster and with fewer defects. Today Diffblue serves four of the 10 largest U.S. banks and several other members of the Forbes Global 2000; customers include Citi, ING, Cisco, AstraZeneca, and BNY Mellon. The company is led by CEO Toffer Winslow. Diffblue intends to use the new funding to expand operations and accelerate development of its product. Diffblue provides autonomous AI-for-code software that helps software teams improve code quality, expand test coverage and increase productivity. The company, led by CEO Mathew Lodge, offers a newly launched version that can autonomously write and maintain unit test suites for entire applications and automatically refactor Java code to improve testability. Its solution also slashes the time and cost to run tests in continuous integration and helps development managers understand test coverage and code risk across their organization. Diffblue raised $8M in funding and intends to use the proceeds to expand operations and broaden its business reach. The company is based in Oxford, UK. Diffblue is a University of Oxford spin-out that applies artificial intelligence to software development, built on ten years of academic research. Its core AI builds an exact mathematical model of a codebase from a few examples to achieve deep semantic understanding of programs. Diffblue is initially developing three products on that engine: an automated testing tool that spots bugs and writes tests, a security product that flags exploitable bugs and generates tests, and a refactoring product that rewrites outdated or poorly expressed code. Founder Daniel Kroening, a Professor of Computer Science at Oxford, says the longer-term vision is for the system to be able to write entire programs from examples. Financially, the company recently raised $22 million in a Series A led by Goldman Sachs Principal Strategic Investments, one year after leaving academia. Diffblue’s stated thesis is that AI augmentation or replacement of software developers is necessary to meet growing demand for software.

Total raised

$38M

Funding rounds

4

Latest round

Grant

Latest activity

Mar 2025