CyberAgent Engineer Details 100% Coding Agent Adoption Across 150-Engineer Org
The talk is a rare case study from game development, where coding agent use has been considered difficult, and it reports the specific organizational steps and the roughly one-year result rather than the tooling itself.
Reporting from 1 source: GameBusiness.jp.
A CEDEC2026 talk by Yusuke Sugiura of CyberAgent and Applibot described how coding agents became the norm for engineers at Applibot, an organization of about 600 people with roughly 150 engineers. A five-person Generative AI Office launched in May 2025 ran hands-on sessions and published a whitelist of approved AI tools to move past study sessions alone.
Applibot runs more than 10 projects in parallel, from teams of over 100 people to teams of around 10, with Unity on the client side and Go on the backend. That scale is what made broad adoption hard to force.
The company stood up a cross-organizational Generative AI Office of about five people in May 2025 to spread coding agents. At that point, most staff did not know about Cursor, and many engineers did not know it existed.
The office clarified available AI tools through a whitelist and built in-house integrations for MCP with tools already in use at the company, including Confluence and Wrike, while running hands-on Cursor sessions open to all job types.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.