From the filing
PDF page 230
No printed folio visible · Image transcription
Read the page text
[Embedded graphic 1 — OCR text] Rakuten Customer since 2024 Turning every employee into a builder Results Business units using Al 70+ Code accuracy delivered on complex code modifications 99.97 Autonomous coding 7+ hrs Working days for new feature time to market 24>5 days Background Rakuten Group is a global internet and e-commerce company operating more than 70 businesses across 30 countries and regions, serving over 2 billion members worldwide. In 2023, Rakuten launched a company-wide effort to build Al into every part of its business, an initiative it calls “Al-nization.” For its developers, this means AI that can build more ambitious products while holding the company’s high standards of quality and security. Impact With Claude Code, Rakuten’s teams could code autonomously for seven hours with limited human guidance, working across complex, multi-language codebases at 99.9% accuracy. It has cut time to market for new features by more than 75%, and Rakuten’s teams now ship major releases every two weeks instead of quarterly. As Claude Code has spread across the company, it has reshaped Rakuten’s teams: roles once split across engineering, design, and reliability have converged, and generalists now take product development from idea to deployment. “Agents like Claude Code shine when we work with them to minimize the cost of new innovation —a quick transition from idea to production. Give a capable person agents that hold context and taste, and it lets hidden talent scale their potential 100 times over.” Yusuke Kaji, General Manager, Al for Business at Rakuten Next, Rakuten deployed Claude Managed Agents across the organization, with specialized agents in product, sales, marketing, and finance, each deployed within a week. Employees give tasks to Claude in Slack or Microsoft Teams and receive finished work, from proposal decks and spreadsheets to working applications. Managers work alongside Claude to clear reviews and triage process bottlenecks. As Claude’s capabilities have expanded, Rakuten’s agents have taken on more autonomy and now complete nuanced, long-running tasks overnight. Rakuten measures every job on task completion and cost per task, routing each to the model where the capability justifies the cost: smaller Claude models handle routine work, while the most demanding tasks run on frontier models at full reasoning effort. “At the highest effort, Fable reflects on and validates its own work. For us, that’s what makes highly autonomous operations possible—the extra thinking pays for itself? Kaji said.
This page includes text transcribed from an image. Check the original for tables, images, and exact formatting.