How Money Forward Uses Claude Code to Cut Engineering Time by 70%

Money Forward serves over 17 million individual users and 400,000 businesses with cloud-based financial software, operating at a scale that demands high engineering velocity. Facing intensifying competition and a strategic shift toward AI transformation, the company recognized that standard developer productivity tools were insufficient.

The core engineering challenge was contextual: previous AI coding tools struggled to maintain coherent understanding across multiple projects, programming languages, and complex financial domain logic. Engineers were not realizing meaningful time savings, and onboarding new developers remained a week-long process.

Money Forward launched the MEPAR (Money Forward Engineering Productivity AI Research) program, structuring Claude Code integration into pull request workflows, developer onboarding sequences, and incident response processes. The team also built specialized internal tooling — Sherlock for incident response and Lumos for automated Kubernetes remediation — on top of Claude Code, plus a custom MCP Server developed by a single engineer over three months.

Adoption was exceptional: 80% of engineers began using Claude Code, with 70% integrating it into their daily workflow. API endpoint implementation time fell from two days to five hours — a 70% reduction — and early adopters saved approximately seven hours per week. New engineer onboarding shrank from one week to one day.

The MEPAR program positioned Claude Code as infrastructure-level productivity tooling embedded in the workflows that define how Money Forward engineers write, review, and deploy code at scale.