Lokalog

🌟 What’s new?

The Task progress detail by contributor view (under Analytics → Tasks) now includes 7 new Translation Review metric columns per contributor: 

• Reviewed Segments
• Acceptance Rate
• Accepted Segments
• Edit Buckets (Light/Medium/Heavy/Content Created)
• Average Edit Score, Review Turnaround (hours)
• Average AI Score. 

Table tooltips were also updated for better clarity around edit effort and scoring.

🎯 Why it matters to you?

You get a granular, contributor-level view of review quality — not just task completion. Teams importing data can now surface review signals previously invisible, helping them spot bottlenecks and make smarter resourcing decisions.
 

🌟 What’s new?

The Tasks Time Analytics dashboard now lets you filter data by date — choose between when a task was created or when it was completed. Previously, the dashboard defaulted to creation date only, with no way to switch. This update is live for all customers all pricing plans.

• Creation date​ — view tasks based on when they were created
• ​Completion date​ — view tasks based on when they were actually finished

🎯 Why it matters to you?

Analytics should reflect how your team actually works, and for most teams, what matters is when work gets ​done​, not just when it starts. This update makes task data more accurate and actionable across a range of workflows:

• Payment and billing reconciliation​ — filter by completion date to get the metrics that match how your invoicing works

• In-progress task analysis​ — get insights on completed languages within a partially finished task, without waiting for all languages to wrap up (e.g. 4 out of 5 languages done? Those insights are available now)

🌟 What’s new?

Project snapshots have been overhauled with major performance and usability improvements:

• 83% faster snapshot creation — what previously took over five hours for large projects now completes in under an hour, with individual snapshots averaging under a second
• Async snapshot restoration — restores now run in the background; you'll receive an in-app notification and email when your restore is complete, so no more waiting on-screen
• 10x smaller snapshot size — compressed storage footprint is dramatically reduced across the board

🎯 Why it matters to you?

If you manage large projects, this is a significant unlock. Previously, restoring a snapshot on big projects could fail entirely, requiring direct support from Lokalise — now you can do it yourself, independently and reliably. 

Faster creation also means the automatic snapshots triggered by bulk actions (pseudolocalize, fill text, copy source to target, clear, and more) no longer slow down your workflow.

🌟 What’s new?

The Lokex package, the Go library powering file exchange between your codebase and Lokalise, has been updated to v2.2.0. This release is a major rewrite focused on stability, with significantly improved test coverage and a cleaner code structure throughout. The GitHub Actions for push and pull will soon be updated to take advantage of this new version.

🎯 Why it matters to you?

If you use Go in your stack, you can now rely on a more dependable and maintainable integration with Lokalise. The improved foundation means fewer unexpected issues during file exchange operations, and the upcoming GitHub Actions update will bring these reliability gains directly into your CI/CD localization pipelines — with no extra setup required on your end.

🌟 What’s new?

We've released ExLokaliseTranfer, a new open-source Elixir library that wraps the Lokalise API upload and download endpoints. With just a few lines of code, you can exchange translation files directly from your Elixir or Phoenix application — no need to hand-roll API calls or build your own retry logic from scratch. The library handles errors and edge cases so you don't have to.

🎯 Why it matters to you?

If you build with Elixir or the Phoenix framework, integrating Lokalise into your workflow just got significantly easier. You can automate file exchange as part of your development pipeline without writing complex boilerplate, saving time and reducing the risk of integration errors.

🌟 What’s new?

We’ve upgraded our AI scoring system to a new model based on Claude Sonnet, improving how translation quality is evaluated. This update brings:

• More precise scoring with better alignment between scores and AI-generated comments
• Fewer false positives for more trustworthy results
• Improved handling of glossaries during scoring

We’ve also added additional resilience for Claude usage via Bedrock to ensure more stable performance.

🎯 Why it matters to you?

You can now rely on AI scoring with greater confidence when reviewing translations. With more accurate and consistent signals, it’s easier to decide what needs attention and what’s ready to go, helping you reduce manual review effort and maintain high-quality output.

🌟 What’s new?

Our GitHub Actions for pushing and pulling files have been upgraded to major version 5, bringing several reliability and performance improvements:

• Updated codebase with multiple internal enhancements
• Proper OS platform auto-detection for smoother execution across environments
• Improved handling of edge cases to reduce unexpected failures
• Compiled with the latest Go 1.26 for better performance and security

🎯 Why it matters to you?

This update makes your CI/CD localization workflows more dependable and easier to run across different environments:

• Fewer errors during automation
• More consistent behavior across operating systems
• Stronger overall performance and stability

In short, your automated translation pipelines become more reliable with no extra setup required.

🌟 What’s new?

Our Elixir SDK has been upgraded to version 4.0.0 with several under-the-hood improvements:

• Updated codebase for better stability and maintainability
• Migration to a newer HTTP client for improved performance and compatibility
• Fully rewritten test suite for stronger reliability and consistency

🎯 Why it matters to you?

If you build with Elixir, this update gives you a more dependable integration experience:

• More stable API interactions
• Faster, more efficient requests
• Greater confidence when deploying or scaling localization workflows

Overall, you get a smoother developer experience and a stronger foundation for building localization into your applications.

🌟 What’s new?

We’ve released major updates to our Ruby tooling to make localization workflows smoother for developers:

• Lokalise file manager for Ruby (v7) now includes stability improvements and official support for Ruby 4.
• Lokalise integration for Ruby on Rails (v9) adds support for working with multiple Lokalise projects from a single Rails app.

🎯 Why it matters to you?

If your team builds with Ruby or Rails, these updates make your setup more reliable and flexible:

• Run localization workflows confidently on the latest Ruby versions
• Manage multiple projects without complex workarounds
• Simplify architecture for apps that handle multiple products, brands, or environments

Together, these improvements help you maintain faster, cleaner localization pipelines with less maintenance overhead.

🌟 What’s new?

Custom AI Profiles are officially out of beta and available to all customers on Enterprise and Elite plans. You can now personalize AI translations at scale using your own linguistic data, including:

• Translation Memory

• Project content

• Reviewed translations

Powered by Retrieval-Augmented Generation (RAG), AI automatically pulls your most relevant examples and uses them as context for each translation. This means your brand voice, terminology, and style are applied consistently — without any manual setup or model retraining.

🎯 Why it matters to you?

With Custom AI Profiles, you get:

• Up to 95% ready-to-publish translations

• 90%+ acceptance rates with human-level quality

• Fewer review cycles and faster turnaround

• Lower localization costs

• Always up-to-date context from your latest approved content

In short, your AI translations now sound more like you — automatically.