LangSync

AI-powered localization platform — manage translation strings, auto-translate them with AI, and serve them to your apps over a REST API.

LangSync centralises every translation string in your product, generates translations with AI, and exposes them as i18n-ready JSON over a REST API. It replaces the spreadsheet-and-Slack workflow most teams ship with: one dashboard for your source strings, a glossary that keeps terminology consistent across every AI call, a CLI that syncs your Git repo with the server, and resumable batch jobs that auto-translate everything you haven't filled in by hand.

The mental model

Everything in LangSync hangs off five concepts:

Concept	What it is	Scope
Namespace	An isolated translation workspace — one per app, module, or client project. Has a default language, target languages, and terms.	Organization
Term	One translation key (`welcome.title`, `button.submit`) inside a namespace. Also called a mark.	Namespace
Translation	The actual string for one term in one language.	Term × language
Language	Either a shared language maintained by Norcube or a custom language you define. Languages do nothing until attached to a namespace.	Platform / organization
Glossary	An organization-wide dictionary of brand names and fixed terminology that the AI consults on every translation. One per organization.	Organization

Translations flow from the namespace's default language into its target languages — either typed by a human, suggested by the AI one cell at a time, or filled in bulk by a background job. Your application then fetches the finished strings through the read-only Public API using an API key.

The dashboard, mapped

Open app.norcube.com, pick your organization, and switch to LangSync. The sidebar has three sections:

Translations
- Namespaces — the list of your workspaces, each with its languages, term count, and a translation-progress bar. Opening one shows its terms and the translation editors. See Namespaces and Manage terms.
- Glossary — the organization's terminology dictionary. See Glossaries.
- Manual translation — a free-form translate box for text that isn't a term: paste, translate, copy. See Manual translation.
Access
- Members — per-service access control: who in your organization can see and do what in LangSync. This uses the platform-wide IAM system; see Platform for how roles and permissions work.
- API keys — long-lived credentials for the Public API. See API keys.
Operations
- Audit log — every sensitive action in your organization's LangSync account, filterable by date, action, target, and actor. See Audit log.

There is one more surface that isn't in the dashboard at all: the LangSync CLI (norcube / nrc), which pins a Git repository to one or more namespaces and reconciles local JSON translation files against the server — including AI translation of anything that's missing.

Where the AI fits

Three flavours of AI translation, all powered by the same engine:

Batch auto-translate fills every missing translation across a namespace — triggered from the dashboard's Translate missing dialog, by a bulk import, by creating a term with Translate automatically enabled, or by a sync job. Runs in the background; you get an email when it finishes.
Single-cell suggestion — the sparkle button next to a translation field generates one AI translation for you to review and save. Nothing is stored until you save.
Manual translation — translate arbitrary text live, optionally with the glossary and a custom prompt applied. Never stored.

Every one of these consults your glossary automatically (manual translation has an on-by-default toggle), so brand names and fixed terminology survive every translation. AI calls are rate-limited per organization and per user — see Limits and errors.

A typical setup

For a product team shipping a web app in five languages:

Create a namespace web-app with default language en-US, attach the four target languages (Getting started).
Seed it — bulk-import your existing CSV/XLSX export, or run nrc langsync init in the repo and push your existing JSON files (CLI).
Set up the glossary with your brand names and fixed terminology before running big AI batches.
Auto-translate the gaps, review in the side-by-side editor, and fix what needs fixing.
Wire your app to the Public API, or keep the JSON files in the repo and run nrc langsync sync in CI.

Pricing

LangSync bills per non-empty translated string stored — empty translations don't count, and every term × language combination is counted once. AI translation calls don't carry a separate per-call charge, but they are rate-limited. See Platform → Billing for the current rates.