Lokalog

🌟 What’s new?

We’ve introduced the Lokalise MCP Server, a new way to control Lokalise directly from your AI tools. You can now connect Lokalise to assistants like Claude or ChatGPT and trigger localization actions using natural language — no manual API calls, no digging through docs, no custom scripts.

With the MCP Server, you can ask your AI to update translations, manage keys, or export content, and it will handle the Lokalise API complexity for you.

The Lokalise MCP Server is launching as a closed beta. We’re inviting teams that are already experimenting with AI-powered development to help shape what AI-native localization looks like next.

🎯 Why it matters to you?

If you’re already building in an AI-first workflow, localization no longer has to be the slow, manual step that pulls you out of it.

• Work with Lokalise directly from your AI assistant

• Automate repetitive localization tasks with simple prompts

• Reduce context switching between tools

• Ship multilingual updates faster, with less engineering overhead

What used to take hours (or days) — navigating dashboards, writing scripts, debugging integrations — can now take seconds.

🌟 What’s new?

You now have more control over how AI models are selected in AI Profiles. For each profile, you can choose to:

• Use automatic routing (recommended): We dynamically select the best LLM for translation quality.
• Select a specific LLM: Override routing when legal, compliance, or security requirements mean you must use an approved provider only.

🎯 Why it matters to you?

You get the best of both worlds: the highest possible translation quality through smart routing where it’s allowed, and full control when your organization needs to restrict which AI models can be used. This makes it easier to adopt AI translations even in highly regulated environments.

Good to know
If you want to use a single, fixed LLM across all projects, you can do so by setting that choice in each AI Profile.

🌟 What’s new?

Style Guides are now used as context not only for Pro AI Tasks & Automations, but also for Pro AI Suggestions directly in the Editor.

🎯 Why it matters to you?

You’ll now get AI suggestions that better follow your tone, brand voice, and style rules — right where you work. After A/B tests showed strong improvements for customers with well-defined style guides, we’ve rolled this out broadly so you get more consistent, brand-aligned suggestions without extra setup.

🌟 What’s new?

You now decide how AI behaves in segmented projects. A new option — “Translate all segments (including existing translations, except 100% TM matches)” — gives you control over whether AI translates:

• Only empty segments (new default), or

• All segments, including those with existing translations.

TM rules stay in place: 100% TM matches always override AI, so you never spend AI credits unnecessarily.

🎯 Why it matters to you?

This update prevents accidental overwrites, reduces unnecessary AI spend, and lets you choose exactly when existing translations should or shouldn’t be refreshed. It also gives you clearer, safer workflows when working with large or partially translated content sets.

🌟 What’s new?

The Lokalise pull action for GitHub now supports a new git_sign_commits parameter, allowing you to sign commits created by the action. This is especially helpful for teams that require commit verification as part of their security or compliance workflows.
To enable signing, follow the short setup guide linked in the GitHub discussion.

🎯 Why it matters to you?

You can now meet stricter security requirements and ensure all Lokalise-generated commits are verifiable — without manual intervention

🌟 What’s new?

We’ve introduced automatic re-translation for any AI-generated string that exceeds your character limits. If a translation is too long, Lokalise will now automatically re-run it with Claude (up to two additional attempts) to produce a version that fits the required length.

🎯 Why it matters to you?

You’ll now see over 98% adherence to character limits, up from a 43.5% baseline (with GPT). This means far fewer manual fixes, fewer UI overflows, and cleaner handoff to design and engineering. It also marks the first milestone of our broader re-translation initiative, which will continue to reduce manual QA and improve overall translation quality.

🌟 What’s new?

We’ve expanded our file format support for Web & Mobile projects with the introduction of .xcstrings, Apple’s default localization format for iOS projects. 

🎯 Why it matters to you?

This update helps iOS development teams reduce manual work caused by unsupported file types. It’s a key step in expanding Lokalise’s coverage of Apple localization formats, ensuring smoother workflows and stronger compatibility with native iOS development tools.

🌟 What’s new?

We’ve released Lokalise Glossary Guard, an open-source utility that automatically checks and fixes CSV files before you upload them as glossaries into Lokalise. It detects and resolves common issues like:

• wrong separators
• encoding problems
• duplicate entries
• formatting inconsistencies

All fixes run automatically in under a second.

🎯 Why it matters to you?

If you manage large or frequently updated glossaries, this tool helps you avoid upload errors and tedious manual cleanup. You get clean, consistent glossary files every time — making your localization workflow smoother and more reliable.

🌟 What’s new?

We’ve enhanced our GitHub Actions for file exchange by compiling binaries with signed and attested checksums. This allows anyone using the actions to independently verify that the binaries come from a trusted source and haven’t been tampered with.

🎯 Why it matters to you?

This update provides stronger security and integrity guarantees — especially valuable for teams in highly regulated industries (such as banking) where supply-chain security is a must. You can now rely on a more transparent and verifiable setup for your CI/CD workflows.

🌟 What’s new?

AI-powered translation quality scoring is now fully available to all teams on Growth, Advanced, Enterprise & Elite plans, concluding the AI scoring Beta phase. 

🎯 Why it matters to you?

This feature uses advanced AI to analyze translations and generate quality scores instantly (based on MQM, an industry-standard), helping your team:

• Identify low-quality translations before they go live
• Focus human effort only where it matters, cutting review time and cost
• Improve reviewer productivity with specific issue guidance
• Make confident, data-backed localization decisions
• Unlock low-priority languages that were previously cost-prohibitive