#### **Mission** Mature GitLab's localization technology stack by evolving from external-tool-dependent, segment-based workflows to git-native, agent-driven localization at scale. #### **Vision** * GitLab Duo agents, operating directly through GitLab primitives - issues, merge requests, CI/CD pipelines, and version-controlled instruction files - handle translation execution end to end. * Language-specific resources (style guides, glossaries / termbases, translation rules) live in version-controlled repositories, are reviewable in MRs, and serve as the authoritative quality layer that agents consume at runtime. * The marginal cost of adding a new language becomes the cost of authoring its specs, not rebuilding infrastructure. * This vision enables significantly faster time-to-market and an the necessary scale runway for providing multilingual customer experience. * Program management depends on data: which files have changed, which are in flight, which have been reviewed by a human, which are pending for a given language. Argo holds that (meta)data - per-file, per-language lifecycle state accumulated across repositories and content streams over time. That is its irreplaceable value as the new architecture matures. How that data is surfaced and queried - whether through dashboards, AI-assisted reporting, or direct issue and MR integration - is secondary and will evolve. * Product UI translations continue independently through the established Crowdin sync in [crowdin-translation-sync](https://gitlab.com/gitlab-org/frontend/crowdin-translation-sync) #### **Alignment with GitLab vision** This infrastructure is a working example of GitLab's AI-native operational model: agents handling routine execution, humans reviewing outcomes, all activity traceable in GitLab. It directly supports international expansion and revenue growth.