Agentic localization QA for live websites and web apps: Every language you ship, checked while you sleep.

Loqalit is an agentic localization QA platform. Its crew checks your websites and mobile apps, live and on staging, in every language you ship, scores translation, context and layout bugs with MQM, and routes the fixes into Crowdin or Phrase for you to approve.

Six bugs are hiding in tonight’s pages. Move your light over the city to find them.

All six. The crew finds these on every page you add, every night.

Scout reads the page your users see.

Not the strings in your TMS. The rendered page, in a real browser, with nobody logged in. When a release brings new critical or major issues, your team gets one email, not fifty. These six only show up there:

  • Text that no longer fitsOverflow and truncation, measured in the browser.
  • Strings nobody translatedEnglish left behind on a German page.
  • Broken placeholdersA raw {count} where a number belongs.
  • The wrong languageLoqalit reads what is on screen, not the lang tag.
  • Glossary termsYour term list, enforced on every page.
  • Formats that don’t matchPrices, dates and numbers in the wrong style.

One email when a release breaks something

When a release brings new Critical or Major issues, Loqalit notices on its own and sends your team one email a day at most, with the page, the language and an example.

Then Scout opens your app.

Your iOS and Android apps get the same night shift: their screens, in every language you ship, with the same six checks. Right-to-left layouts included. Every finding comes with an MQM severity and a screenshot, and accepted fixes go to Crowdin or Phrase.

Mobile app localization QA →

9:41
Konto
LF
Lena Fischerlena@beispiel.de
  • Bestellungen
  • Account settings3
  • Zahlungsarten
{count} Artikel im Warenkorb1
Jetzt kostenlos registrieren2
de-DE
9:41
الحساب4
ل
ليلى أحمدlayla@example.com
  • الطلبات
  • الإعدادات
  • طرق الدفع
٣ منتجات في السلة
تسجيل الدخول
ar
Tonight’s app scan4 findings · MQM-scored
  • 1Critical

    Raw {count} in the cart line

    de-DE
  • 2Major

    Sign-up button text is clipped

    de-DE
  • 3Major

    “Account settings” left in English

    de-DE
  • 4Minor

    Back arrow not mirrored

    ar

Editor makes it fit.

It measures the space on screen, checks your glossary and drafts wording that fits. Then it waits. Editor never ships anything on its own.

Jetzt kostenpflichtig bestellen
Text needs296 px
Button has230 px
Languagede-DE
  • Fits at every width down to 320 px
  • Keeps your glossary term “bestellen”
  • Keeps the legally required “kostenpflichtig”
Watch it rewrite the label.

Verifier checks twice.

A fix only counts after two clean scans, six hours apart. If the bug comes back with a later release, it goes straight back to your inbox.

FIXED
06:00Fix is live
06:10Clean scan one
12:10Clean scan two

Regressions get the same treatment: a fixed bug that returns is reopened, not forgotten.

You say yes over coffee.

Approve or reject what the crew drafted overnight, then decide when it works next: daily, weekly, quarterly or once, inside your quiet hours and a budget you set. Accepted fixes land in Crowdin or Phrase as a comment with a screenshot.

See how it works on live websites and web apps →

Set the cadence once. The crew shows up on time.

Your live and staging pages get scanned daily, weekly, quarterly or once, in the languages you choose, with nobody at the keyboard.

  • Runs in your quiet hours
  • Live and staging, in a real browser
  • A budget cap that ends scans cleanly

If a language breaks, the release waits.

Add Loqalit as a step in GitHub Actions. If a scan finds critical or major issues, the build fails and the release waits.

  • Runs with your own API keys, after the deploy to staging
  • Every failure names the page and the language, with a screenshot
  • Not on GitHub? Signed webhooks and a REST API
release/2026.10.3GitHub ActionsWaiting
  1. Build
  2. Deploy to staging
  3. Loqalit localization check
  4. Deploy to production
An example run

Your translators never leave your TMS.

When a translation job is done, Loqalit scans that language and leaves each bug as a comment on the string, with a screenshot. Nobody opens Loqalit.

  • Every string shows where it sits on the page
  • Dry run first, and no comment when it isn’t sure
  • Your other tools hear when a scan finishes

We ship languages. reviewers spend about hours per language each month hunting for bugs on screens. That’s 2,592 hours a year.

Loqalit’s crew does the hunting, and % of that time goes back to your reviewers: 1,555 hours for tone, terminology and cultural fit.

Worth about $41,990 a year at $ an hour.

Edit any underlined number. The hours are in-context review: opening each page and screen in each language to look for bugs. The share is the part of that review Loqalit’s checks can take on. An estimate, not a quote.

That’s the time checking takes. Now the cost of missing one.

If a broken checkout loses % of $k in monthly non-English revenue for days before a customer reports it, that’s $5,833 gone from one bug.

Caught by the next nightly scan, it’s $417.

Monthly revenue × (days ÷ 30) × conversion drop. The second figure uses one day: it assumes the bug turns up in the next nightly scan. An illustrative scenario: change the numbers to yours.

New

Ask in the chat. The crew answers.

Connect Loqalit to Claude, ChatGPT or Cursor. Ask in plain words to check a live page or a set of strings, and the findings come back in the conversation, with MQM severities and suggested fixes.

  • One server address, sign in once
  • Also in Claude Code and other MCP clients
  • New accounts get 3 free AI analyses
Connect your assistant

Hand it to the crew, or check a page yourself.

The Loqalit platform

The crew scans your live and staging pages on your schedule, tracks every defect across releases, and drafts fixes for you to approve.

Starts with a free QA report of your own site and a demo.

The Chrome extension

Open the side panel on any page, including pages behind a login, and see each finding next to the text it came from.

A side tool for checking one page on the spot. One free guest scan per device.

Built for localization managers, product teams, QA engineers and translation agencies who would rather find the bug before a customer does.

Before you log off.

Frequently asked questions

Quick answers about Loqalit, pricing, and localization QA. Anything else? Write to info@loqalit.com or request a demo.

What is Loqalit?

Loqalit is an agentic localization QA platform that checks your live and staging websites, web apps and mobile apps in every language you ship. It catches translation, context and layout bugs in the rendered product, scores them with MQM, and routes fixes into Crowdin or Phrase.

Is Loqalit a Chrome extension?

No. Loqalit is an agentic localization QA platform: its agents run scheduled scans of your pages with nobody logged in. It starts with a free QA report of your site and a demo, and our team onboards you. A Chrome extension is also available as a side tool for checking a single page on the spot, and anyone can try it today.

Why use Loqalit instead of pasting text into generic AI chats?

Pasting strings into a chatbot loses the one thing that matters most: context. The AI never sees your rendered UI, so it can’t catch overflow, truncation, layout breaks, or translations that are technically correct but wrong in place. Loqalit scans your live product at the DOM level, scores each issue against MQM, and shows you exactly where it appears.

Which pages and sites can I scan?

Any live or staging web page, including single-page apps and Shadow DOM components; with the extension, that includes pages behind a login. Cross-origin iframes such as third-party chat, payment or consent widgets can’t be read, and text drawn into a canvas is flagged for a human check instead of being skipped silently.

Can Loqalit check mobile apps?

Yes, on iOS and Android. Loqalit checks your app’s screens in every language, with the same checks it runs on websites and web apps, right-to-left layouts included. Findings are scored with MQM, and accepted fixes go to Crowdin or Phrase. See mobile app localization QA.

How does the AI detect issues?

In two layers. Instant rule-based checks for placeholders, number formats and required disclosure text run on every scan without using an AI analysis. An AI analysis then reviews the rendered text and returns each issue with its MQM category, severity, the string and an explanation. On the platform, three agents with separate permissions find new pages (Scout), recheck fixes over time (Verifier) and draft wording (Editor), and every change they suggest needs a person’s approval.

Is a trial available before subscribing?

Yes, for the Chrome extension. Try the Chrome extension as a guest: one free scan per device, no account needed. For the platform, start with a free QA report of your own site.

Does Loqalit work for teams and enterprises?

Yes. Teams share a workspace with roles and a shared scan history, and our team onboards you after a demo.

What is localization QA?

Localization QA is the process of checking translated content in the product it ships in, rather than in a file. It covers linguistic accuracy and fluency, terminology and locale conventions, and the design dimension: text overflow, truncation, and layout breaks caused by translated text. A translation can be correct as a sentence and still be a defect in the interface it lands in.

What is MQM scoring?

MQM (Multidimensional Quality Metrics) is a framework for classifying translation defects by error type and severity. Errors are grouped into dimensions such as accuracy, fluency, terminology, style, locale conventions, and design, and each is assigned a severity, typically critical, major, or minor. A score is calculated by subtracting weighted penalty points from 100, normalized against sample size. Loqalit scores every finding against this framework. See our full guide to MQM scoring on the blog for the complete error typology.

Does Loqalit replace my TMS?

No. Your TMS remains the system of record for translation, and its built-in QA checks are worth keeping switched on. Loqalit is the verification layer between translated and shipped: your TMS checks the text you sent it, Loqalit checks the product your users get. Loqalit integrates with Crowdin and Phrase.

Do I need to install anything to use Loqalit?

No. Set up your pages, languages, and a schedule once, and Loqalit’s agents take it from there, with no login and no extension required for scheduled scans. A Chrome extension is also available when you want to scan a page on the spot. Access to the platform starts with a free QA report and a demo.

How much does Loqalit cost?

The agentic platform starts with a free QA report and a demo. Guests can try the extension with one free scan per device, no account needed.

Can Loqalit detect text overflow and layout bugs?

Yes. Because Loqalit evaluates the rendered page at the DOM level rather than a string list, it detects defects that only exist after rendering (text overflow, truncation, clipping, and layout breaks introduced by translated text) alongside linguistic and terminology issues, in the same scan.

Which languages does Loqalit support?

Loqalit is not restricted to a fixed list of locales. Because it analyses the text rendered on the page rather than importing from a predefined language set, it works with virtually any language your product ships in, including non-Latin scripts and right-to-left languages such as Arabic.

Schedule it. Forget it. Ship it.

Start with a free QA report. We scan your live site in up to 3 languages and email you the findings with screenshots, MQM severities and suggested fixes.

No install, no credit card. Rather talk first? Request a demo

Set the cadence once. The crew shows up on time.

Your live and staging pages get scanned daily, weekly, quarterly or once, in the languages you choose, with nobody at the keyboard.

  • Runs in your quiet hours
  • Live and staging, in a real browser
  • A budget cap that ends scans cleanly