Zapier uses Codemod to turn years of tech debt into a background task
The critical project no one could afford to start
Over the years Zapier had collected thousands of hard-coded strings across many repos, owned by hundreds of developers. They wanted to extract them so the Content Design team could control messaging and ship updates without involving engineers, but the manual work made it too tedious and too expensive to ever make the roadmap.
When generic AI hits real-world code, things fall apart
Generic AI agents are great at generating new code, but they run into real limits when asked to remediate code across a large, complex production environment. With tight context windows, low repeatability, and slow execution across large repos, they couldn’t reliably detect and extract strings across 70M+ lines of production code. And even if they could, Zapier still faced an organizational challenge: planning, tracking, and coordinating changes across teams and repos. No tool solved both the technical and operational sides of the problem.
The backlog finally got too big to ignore
As Zapier’s engineering footprint grew, the team needed a better way for Content Design to control in-product content at scale. Through Lokalise, they were introduced to Codemod as the solution that could support this safely across their codebase.
Building a localization-ready engineering environment that can scale
Zapier’s 350+ developers work across a large, multi-repo GitLab codebase with Jira as the planning system. Lokalise serves as their Translation Management System, giving them a unified translation workspace tightly integrated with development and deployment. Features like translation memory, glossaries, version control, and in-context editing turned content updates from a manual bottleneck into an automated, developer-friendly workflow. With that foundation in place, Codemod became the natural next step automating the code changes needed for string extraction at their scale.
Choosing the only platform built for both code and coordination
Zapier chose Codemod for three reasons:
- i18n micro-agents customized for Zapier’s codebases
- Leadership dashboards for planning, tracking, and validating progress
- Campaign orchestration to coordinate safe rollouts across teams and repos
They also saw Codemod as a long-term partner for future code maintenance and remediation efforts, from design system upgrades to enforcing best practices, strengthening AI guardrails, and security fixes.
One engineer pressed go — automation did the rest
A single campaign owner pointed Codemod at the target repos, configured the i18n pro codemod with pre/post-run scripts, AI review prompts, and batching rules, and let automation take over. Codemod created Jira tasks and high-quality PRs within minutes. Weekly check-ins with Zapier’s tech lead ensured quick feedback loops on Codemod behavior and platform improvements.
Codemod Campaign, a configurable workflow for multi step transformations tailored to the Zapier codebase
A project stuck for years is now moving in the background
- 90%+ automation coverage
- Months of manual rewriting reduced to a background task
- One engineer running work that previously demanded ten
What progress finally looks like
- Campaign view for the i18n pro codemod with nodes and steps configured by the campaign owner
- Automation coverage dashboard
Codemod Insights time series widget showing a burndown chart of hard-coded strings
The business finally gets enterprise polish without slowing product
String extraction now has momentum. Leadership gets a faster path to content moderation, while engineering teams stay focused on building product instead of grinding through repetitive refactors.
See what your engineering org could automate next
For engineering leaders: Schedule a demo and request a pilot to explore Codemod’s task-specific micro-agents for large-scale code maintenance and remediation, including internationalization, post-M&A codebase sync, security remediation, EOL upgrades, framework shifts, vendor SDK migrations, and more.
Put software maintenance on autopilot.
De-risk, automate, and orchestrate large-scale code changes with secure, reliable, and lightning-fast micro-agents—built by your engineers or Codemod experts.