Is there AI behind Localazy?

Learn a bit more about advanced technologies behind Localazy that are making your app localization better while you sleep.

From first sight, Localazy is similar to some other translation platforms. We invest a lot of effort into keeping it as simple as possible while doing everything we can automatically for you in the background. Before adding a single button for a new feature, we discuss it thoroughly, looking for a way to do the same operation without the button - preferably automatically.

Also, there are a lot of differences between Localazy and other translation platforms in the background. In Localazy, there are never-ending processes that constantly go through your app, trying to translate it or suggest better translations wherever possible and improving the translation process.

Some companies would instantly mark this with AI stickers and market the platform as AI-powered. We indeed use mechanisms from machine learning, data mining, and big data processing. However, we mostly rely on complex hand-crafted algorithms designed for massively parallel processing.

For example, for every single translated phrase, thousands (yeah, really) of processes are launched to distribute the new translation across other phrases in your and connected projects (aka InTM) and across the whole database to all other apps (aka ShareTM) via shared translation memory. For ShareTM and suggestions / similar phrases that provide additional context for your translators, internal ratings are pre-calculated, so we can offer the best matching candidates at the speed of light. And every single new translation can change these ratings…

Shared translation memory

So, we are not only storing a few bland lines of texts 😃. There are Oompa-Loompas continually working to make your app better and better. While we all sleep, hundreds of CPU cores at Amazon’s data centers are working hard…

Yet, I wouldn’t call this AI as you are still the one to have the last word and control the things.

However, we aim to ideally remove you from the equation, making translating your mobile app as smooth as uploading it to Localazy… and that would be everything on your side; from that moment on, Localazy would be in charge. We already see some great results from trained machine translation models - they could act as reviewers processing the massive amount of the context information and producing qualified decisions just as humans do.

Translation Memory

Machine translations are becoming better and better, but without the context and trained app-specific models, they still have limitations.

It’s where our journey starts!