Detection — it looks at what your users look at.
Scans the rendered live page in a real browser, not the translation file — so it sees what the customer sees.
Every target language on the project.
MQM Core classification: 25 categories, severity-weighted into a 0–100 score per page.
Every finding carries a 2x screenshot of the element where it sits, the reason it is wrong, and a suggested fix.
New to rendered-product QA? Read the guide: Localization QA for live websites and web apps.
Only what changed — the Inbox shows the delta, not the backlog.
Findings keep an identity across runs: new, regressed, or fixed.
The Inbox shows the delta, not the backlog.
Dismiss a finding once and it never comes back.
Accept or ignore one at a time, or in bulk. A Baseline view for the rest.
Automation — nobody has to be logged in, or awake.
Scheduled scans: daily, weekly, quarterly, or one date in the future.
Runs are dispatched server-side. Nobody has to be logged in, or awake.
Post-publish drift: Loqalit fingerprints every page it sees, notices a republish, and works out what that release broke.
Email alerts for regressions and drift, with per-person email preferences.
The agents — everything they do arrives as a proposal.
Scout crawls from one page you gave it, reads the sitemap, and proposes the pages worth watching. It can’t add them to your project itself.
Editor reads the exact string out of the scan and drafts a correction that fits the space and respects your glossary. It never writes to your site.
Verifier resizes the browser, measures the element, photographs it and checks the glossary — and cannot change a finding’s status. Its two rechecks, six hours apart, are what an issue needs before it’s called fixed.
Ask any of them in plain language, by typing or with your voice.
Every action is a proposal a person reviews on a card, and nothing moves until someone presses Confirm.
Your standards, enforced — checked against your rules, not a generic bar.
Glossary import and pinned terms.
Brand voice and QA style guide per project.
Mute categories you do not want reported.
A pass threshold per project, from 50 to 99.
Delivery and integrations — the fix lands where your team already works.
Accepted corrections posted server-side to Crowdin or Phrase as comments with a screenshot.
Slack: after a scan, you can send its summary to your Slack channel.
Signed webhooks and a REST API with your own keys.
A Chrome extension for scanning any page on the spot.
Reporting — every run, every finding, on record.
Export to PDF, XLSX, CSV and JSON.
Language coverage and its trend over time, so a drop can be dated.
Per-page history: every run, what is open, fixed or dismissed.
An activity log of every run and every step an agent took.
Team — one workspace for everyone on the project.
Workspaces and projects.
Seats and roles.
Invitations.