Capcom’s REX Project Targets RE ENGINE’s Growing Bottlenecks
REX targets long development cycles and RE ENGINE's learning curve through data, tool and runtime work, with AI-assisted coding and testing among Capcom's longer-term goals.
Capcom’s REX project aims to ease growing development bottlenecks in RE ENGINE without replacing the engine. At CAPCOM Open Conference RE:2026 on Oct. 2, programmer Satoshi Ishida described two challenges behind the effort: major games take too long to develop, and staff without experience using RE ENGINE can find its tools difficult to learn. REX brings together work on data handling, development tools and runtime performance, alongside longer-term plans for AI-assisted development, as outlined in coverage of the presentation.
Capcom began developing RE ENGINE in 2014, and the engine first appeared in a commercial release with Resident Evil 7 in 2017. It has since powered more than 27 Capcom titles and is used by over 2,000 developers across countries and technical backgrounds. As games become more complex and data-heavy, routine changes can take longer and development tools face greater demands.
REX targets data, tools and runtime work
RE:Dox aims to standardize how RE ENGINE handles different kinds of game data, with the goal of improving loading and editing speeds. RE:UI is intended to use AI to optimize Capcom’s development software, reducing unnecessary memory use and processing while helping interfaces stay responsive as the tools grow more complex.
RE:Log changes how the development environment stores and retrieves logs. Rather than repeatedly handling large blocks of text, it can use compact identifiers to call up relevant information when needed. Capcom says the approach reduces the amount of data being handled and makes logs easier to search, analyze, edit and translate. RE:Dox and RE:Log are already available as open-source projects.
RE:Flows focuses on gameplay logic. Designers can build that logic visually, and the system translates their work into a standardized programming language. This can help teams implement gameplay ideas without manually coding every element, but the system translates designer-authored logic rather than inventing mechanics on its own.
RE:Runtime is aimed at performance when many characters and objects are active at once. It is designed to animate groups together instead of processing each real-time animation separately, while managing memory and processing more efficiently. Alongside it, RE:C++ translates developer-friendly code into a form optimized for faster machine processing.
AI remains a development goal
Capcom sees a shared programming foundation as a way for people and AI systems to work with the same development tools. Longer-term possibilities include AI assistance with coding, generating portions of a game during production, running tests and playing builds to find bugs. Capcom has not announced a timetable for these capabilities. Ishida also said the company plans to make some technologies available online for people to use in AI training.
Capcom has said it will not use AI-generated assets in its games, keeping that policy distinct from its plans for development support. Shinichi Inoue, the company’s VP of game development platform and AI solutions, said Capcom was not using AI to create art and described the broader goal as helping unlock creators’ potential, as VGC reported. Capcom’s position on AI-generated assets provides further context.
Which problem should REX help Capcom’s teams solve first, long production cycles or the learning curve for developers new to RE ENGINE? With RE:Dox and RE:Log already open source, stay connected for updates on the tools and REX’s development on X, Bluesky, YouTube, Instagram, Steam and Telegram.
