CyberAgent accelerates with ChatGPT Enterprise and Codex | Keryc
CyberAgent, the large Japanese advertising, media and video-game company, decided to make artificial intelligence something everyday, not an experiment hidden in a lab. The result? Employees who use AI with confidence, faster processes, and adoption that already reaches 93% of the workforce each month.
Want to know how they turned AI into a regular tool and not a guarded secret?
What CyberAgent did
The strategy was simple but deliberate: integrate ChatGPT Enterprise and Codex as central pieces of the work environment, with security and clear rules around use. They didn’t force tools by decree; teams and subsidiaries evaluated and adopted them according to their needs. Still, usage grew to a monthly active user rate of 93%.
They've been investing in AI for years. In 2016 they created AI Lab and in 2023 launched AI Operations Office to operationalize the technology. Even before the generative AI wave, in 2020 they released Kiwami Prediction AI to improve creative processes in advertising.
How it changed daily work
With ChatGPT Enterprise, routine tasks like research, drafting and organizing ideas became standard practice. The key wasn’t automating final decisions: humans still make the important calls, but now they get faster drafts and preliminary analyses.
What about safety? They applied corporate controls: account management, visibility into usage, and internal policies on confidential information. That answered the biggest question slowing adoption: what can I put into these tools and what should I avoid?
"With enterprise features like account management and usage visibility, it was possible to support broad use of information while excluding confidential data," says Ken Takao, Manager in the Data Technology area.
Codex: more than a code generator
Codex proved useful beyond writing functions. They use it to:
Evaluate and test design proposals from multiple perspectives.
Suggest improvements during code reviews and offer options.
Keep documentation alive, for example AGENTS.md, which helps agents retain context.
Concrete benefits:
Better design quality thanks to varied proposals.
Faster alignment before implementation, which reduces rework.
Clearer reasoning that speeds decision-making.
Not only developers use Codex: non-technical roles employ it for specifications, mockups and structuring product-adjacent work. A concrete example: the team used Codex to build their internal usage ranking system, which helped make adoption visible.
Sou Yoshihara, in AI Business Division, says Codex produces high-quality proposals and helps improve the overall development process. In game development, Hidekazu Hora reports that Codex helped WormEscape reach soft launch in about a month.
Culture, training and governance
Adoption wasn’t only technological: it was cultural. Sharing prompts, success cases and personal rankings (visible only to each employee) encouraged curiosity instead of fear. When someone wasn’t using the tools, a Slack bot asked why and offered help or alternatives.
Training was intense: OpenAI and CyberAgent organized more than ten sessions with 100+ participants each, ranging from introductions to workshops on custom GPTs, hands-on Codex sessions and internal hackathons. Those practical experiences made early wins easier and cemented ongoing use.
Results and lessons
Massive adoption: 93% monthly active users with ChatGPT Enterprise.
Greater speed and quality in design and implementation, with Codex supporting early-stage decisions.
Effective governance that allowed trust in the tool without sacrificing security.
CyberAgent doesn’t see AI as a fad. They see it as the next standard of the internet and as an engine to redesign how work is done: teams choose, test and scale what works without rigid mandates.
The lesson for any company is clear: combine powerful technology with practical rules, training and a culture that promotes learning, and you can turn AI tools into real work partners. Ready to think about how to apply this in your team?
