Application Refactoring
Stripe needed a CLI-native coding assistant that met enterprise security requirements and could be safely distributed to engineers without exposing corporate laptops to unapproved JavaScript packages or an unmanaged npm dependency chain.
Reported impact
Stripe deployed Claude Code to 1,370 engineers through a zero-configuration enterprise rollout. One team migrated 10,000 lines of Scala to Java in four days, work estimated at ten engineering weeks by hand, enabling an upgrade to a newer JDK. Engineers reported higher satisfaction with their tooling, and Stripe is exploring agent-based incident detection and resolution.
Stripe's developer infrastructure team chose Claude Code for engineers who prefer a command-line workflow and worked with Anthropic over two to three months to produce a signed enterprise binary. The binary bypasses Claude Code's npm dependency chain, addressing concerns about supply-chain attacks from random JavaScript packages on corporate laptops. Stripe distributes it through its internal mechanism, pre-installed on laptops and development boxes and pre-configured with rules, tokens, and authentication, so engineers can use it without self-service accounts or setup. Engineers supply repository and documentation context; Claude Code understands the codebase, edits files, and runs commands while developers review and adapt outputs. The article reports a concrete Scala-to-Java migration using Claude models and describes training, all-hands, and team-level sharing to improve use. The story does not specify a separate data store or an automated deployment flow.
Stripe engineers use pre-installed, pre-configured Claude Code with their codebase and internal documentation as context; Claude Code edits files and runs commands, and engineers review resulting changes. For a Scala-to-Java migration, Claude models converted 10,000 lines and enabled an upgrade to a newer JDK.
Deployed Claude Code to 1,370 engineers with a zero-configuration setup that works immediately on every developer's machine
It's pre-installed on everyone's laptop. It's pre-installed on everyone's development box. It's pre-configured with the rules, the tokens, the authentication.
Migrated 10,000 lines of Scala to Java in four days, a project estimated at ten engineering weeks without AI assistance
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Anthropic Sample Archived · Feb 6, 2026 Figma Figma wanted teams to turn ideas and static designs into interactive software much faster, reduce the friction between design and development, and let non-coders prototype without rebuilding designs by hand. I spent 30 hours teaching myself React to build my website. With Claude, I refactored and improved all that… Claude Opus 4.6 · Claude Platform · Claude Sonnet 
OpenAI Sample Oct 9, 2026 Asana Asana needed to cut the cost and latency of its StackAI browser agent so it could offer customers more capable models for browser-driven workflows without unsustainable operating expense. $0.47 average estimated model cost with the optimized GPT-6.1 Sol workflow Codex · Command · GPT-6 Astra OpenAI Sample Oct 8, 2026 Braintrust Braintrust wanted to respond to customer feature requests faster and run more engineering experiments, but slower coding tools and backlog-driven workflows made it costly to turn feedback into working prototypes in real… 50% of the Braintrust team moved to Codex in one month Codex · GPT-5.5 