Capcom is exploring AI agents to test game builds and flag bugs, while saying AI will not replace people or generate assets for its games. The plan puts AI to work on repetitive development checks, with human developers still responsible for deciding what fixes to make.
Watch Desk analysis
What happened
At an industry conference on 2 October, Capcom discussed how it wants to develop its RE Engine and use AI in game development, Kotaku reports. The company is making two tools, RE:Dox and RE:Log, available as open source. Capcom hopes AI systems will learn how those tools work and help inspect the company’s code.
The proposed agents could play game builds, log issues and look for visual, gameplay, memory or performance bugs, then suggest fixes for developers to implement. Capcom says the aim is to reduce repetitive work and give people more time for creative tasks. It also said AI would not replace people, and has previously said it does not plan to include AI-generated assets in its games.
Why it matters
This is a concrete, behind-the-scenes use for AI in game development: testing builds and helping find problems, rather than generating characters or scenery. If it works as Capcom hopes, it could help teams spot issues across complex games and hardware without handing an agent the final say over a fix.
The distinction is important. The company described an intended use, not a demonstrated reduction in bugs or development time. And a suggested fix still needs a developer to judge whether it is correct.
Our read
AI that helps find a bug, then leaves the repair to a person, is a more grounded pitch than the usual promise to automate creativity. The test will be whether these tools find useful problems reliably, not whether the demo makes a tidy slide.
What to watch
- Whether Capcom publishes the RE:Dox and RE:Log tools and documentation.
- Whether the proposed agents make it into its development workflow.
- Whether Capcom shares evidence of bugs found or time saved.
Discussion spark: Should game studios use AI agents to test builds and suggest fixes, provided developers make the final call, or does that still put too much trust in automated code analysis?
Sources and evidence
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.