Among the use cases GitLab highlights is one labeled “Translate Application to Another Language.” Despite the name, the example is not about translation. Instead, it focuses on helping developers prepare applications for localization by addressing common internationalization challenges.
Localization Readiness
In practice, many localization delays and cost overruns are not caused by translation itself, but by insufficient internationalization. Common issues include hardcoded strings embedded in code, missing support for plural forms, incorrect date and number formatting, and layouts that break in right-to-left (RTL) languages.
Fixing these problems often requires time-consuming engineering work spread across large codebases. As a result, internationalization is frequently postponed or handled late in the development cycle, creating downstream friction for localization teams and Language Solutions Integrators (LSIs).
GitLab’s approach targets this exact problem: the upstream engineering work that determines whether localization can proceed smoothly in the first place. In this use case, the AI agent is prompted to help developers clean up and prepare code by, for example:
- extracting hardcoded strings into translation files
- setting up or standardizing internationalization frameworks
- ensuring locale-aware date and number formatting
- supporting RTL languages
- organizing and maintaining translation resources
Rather than translating content, the AI assists with the mechanical, repetitive engineering work that makes localization possible.
GitLab’s AI agent strategy also extends to the localization side. Separate from the internationalization use case, GitLab’s localization team is currently using AI agents within GitLab Duo to automate translation workflows for marketing website content, including terminology management, style guide enforcement, and automatic merge request creation. Together, the examples show AI agents being used both upstream (internationalization) and downstream (localization workflows).
