Rome – September 11, 2025
Translated, a global leader in AI-powered language solutions, announced today a new release of Lara Translate, making it the first AI translator that users can truly customize. Unlike traditional translation tools and platforms, which rely on static glossaries or offline customization, Lara Translate now learns from user edits, adapting translations in real time. Users simply click the “‘Add to MemoryLearn’” button to save their preferred choices into the new Personal Memory, a user-specific database that Lara relies on to adapt future translations.
Teaching Lara Your Preferences
Personal Memory works hand in hand with the new feature, Alternative Suggestions, enabling users to actively ‘teach’ Lara how to translate the way they prefer, on the go. Alternative Suggestions lets users request different translations for specific words or phrases by clicking on them. Lara provides multiple alternatives and explains why each may be appropriate. When users confirm a choice or refine it further, they can choose to save that preference to Personal Memory to instruct the model on the user's preferences.
This new capability builds on standard personalization tools such as translation memories (TMs, databases of past translations) and glossaries (approved translations for key terms), and is enhanced by contextual instruction, where users can guide Lara with additional context to refine translation choices. It is then further refined through Lara’s translation styles (Faithful, Fluid, and Creative), which adapt tone and register to each use case, from technical documents to marketing campaigns. The result is a translator that no longer produces the same output for everyone, but one that continuously adapts to individual preferences and team requirements
This combination of edits, alternatives, contextual instruction, and styles is the first of its kind in AI translation, setting Lara apart from tools that still rely solely on traditional translation memories and glossaries.
Team-Level Personalization
Teams now have access to Account Memory, a shared resource that all members can contribute to. The team’s admin manages it, reviewing or removing entries when necessary. This setup ensures collaboration across team members while keeping translations consistent and centrally managed.
Seamlessly Working With Existing TMs
Personal Memory is designed to complement, not replace, existing translation memories. Traditional TMs remain a reliable reference, while Personal Memory evolves with each saved edit. When both are active, Lara seamlessly merges their suggestions, preserving consistency while adapting to user preferences over time.