Managing localization at scale often stalls product releases. The traditional workflow requires manually defining keys, bouncing text between design tools and code, and waiting for translations before a feature can go to production.

In this talk, we’ll share how we flipped this workflow upside down at Holafly, making our engineering localization cycle much faster.

We will walk through the technical architecture of this code-first approach, including:

- How a custom localization skill makes it simple to add new texts.
- How we use the Model Context Protocol (MCP) to give the AI visual layout context.
- How we mitigate layout risks like text expansion.
- How to scale your Flutter app to new languages without overhead through a localization CI/CD pipeline.

You'll leave this session with practical strategies to eliminate manual copy-pasting and automate your localization pipeline using AI development tools.