CyberAgent reports massive use of ChatGPT Enterprise and clear improvements with Codex. What does it mean that 93% of their monthly active users use the tool? Simply: it's no longer an experiment — it's part of the daily workflow.
Key results
The stat that stands out is the 93% monthly active use of ChatGPT Enterprise. That tells you the platform isn't just satisfying curiosity; it's woven into everyday design and development tasks.
Takao, the project lead, highlights three concrete use cases where the tool adds value:
- Reviewing, evaluating and stress-testing design proposals from multiple perspectives.
- Generating improvement suggestions during code reviews and offering several options.
- Helping choose between alternatives with clear arguments that make decision-making easier.
Benefits observed with Codex
Beyond use cases, CyberAgent identifies three concrete advantages from adding Codex into their workflow:
- Higher design quality: proposals arrive with different approaches the team can evaluate, raising the overall standard.
- Faster alignment before implementation: less rework because decisions are refined before code is written.
- Clearer reasoning: suggestions include justification that speeds up decision-making within the team.
Practical outcome: fewer back-and-forths, and more informed decisions early on.
Why this matters for your team?
If you work on digital products or development, CyberAgent's experience is a real-world example: AI can boost both creativity and process discipline. It's not about replacing people — it's about giving you an extra pair of eyes that offers alternatives and explanations.
Think of design reviews where the system suggests angles your team hadn't considered, or code reviews where Codex proposes multiple implementations and their pros and cons. How much time would you save if many discussions were faster and based on concrete options?
Practical recommendations to get started
- Start small: design reviews and code reviews are ideal entry points.
- Measure impact: track rework, decision time and monthly adoption (yes, that 93% metric is key).
- Prioritize explanation: choosing a suggestion is easier when it comes with reasoning.
- Keep human oversight: use AI to contrast and enrich, not to automate decisions without supervision.
The lesson from CyberAgent is simple: when AI is used to explore options and justify proposals, it stops being an experimental toy and becomes part of the company's operational pulse.