Rakuten
Claude Code across engineering teams
Rakuten reports time to market down 79% after adopting Claude Code
Problem
Engineering work such as large codebase changes and feature delivery took long, with a monorepo update estimated at over a month by hand.
Approach
Engineers used Claude Code for tests, API mocks, bug fixes, documentation and code review, often running several sessions in parallel. In one task Claude Code worked autonomously for 7 hours in the 12.5-million-line vLLM library. Non-engineers also got access, and specialist agents for product, sales, marketing and finance were set up.
Results
Takeaway
Anthropic reported a seven-hour autonomous coding run at Rakuten.
Categories
- Size
- Large enterprise
- Industry
- Technology and telecom
- Business function
- IT and R&D
- Innovation type
- Process innovation
- AI technology
- AI agentsGenerative AI
- Adoption
- Vendor solution
- Evidence
- Vendor's case study
Sources
Source summaries
Anthropic describes Rakuten's use of Claude Code and managed agents in engineering and business teams. It reports a 79% cut in time to market and a 7-hour autonomous coding run.
Full details are in the original sources.