Platform Core — Working Editor
Everything on this page runs in your browser and is stored in browser localStorage only. Confirmed today: 150+ languages, 75,000+ native linguists, 150,000+ projects., across 2,852 roster-approved language pairs.
1. Project intake —
Documents first. Websites, software strings, courses, video and design follow one content type at a time, each round-trip verified before it is offered.
Sworn work is signed only by a certified human sworn translator; this page never signs sworn work. A sworn Job needs a named Human Gate decider and a reason before its gate can approve (section 8).
2. Roster-before-sales gate —
The Roster below is anonymised (Linguist A to D only). A pair is sold only when a roster row exists for it; otherwise the project waits and no assignment is possible.
| Pair | Translator | Reviewer | Domain |
|---|---|---|---|
| en to ar-SA | Linguist A | Linguist B | Legal |
| en to es-ES | Linguist B | Linguist C | Marketing |
| en to fr-FR | Linguist C | Linguist D | Technology |
Roster covers 3 pairs only. Every other pair in the intake list — including en to de-DE, en to pt-PT, and the five template pairs — has no Roster row and must show WAIT — NOT SOLD. Assignment also requires the Roster domain to match the project domain — a pair match with a different domain also shows WAIT — NOT SOLD. Reviewer must differ from translator; the Rows above already satisfy that rule, and the assignment check enforces it.
3. File / text import and segment editor —
PDF files are not parsed here (this page reads files in the browser only). PDF conversion is live on the platform's CAT Tool — txt, docx and pdf are three of the 103 formats proven to convert live. A PDF you choose will be listed with that honest note and will add no segments. Nothing is uploaded anywhere — every file is read in your browser only.
Segments only. Editing a target after a QA run resets the QA run flag, so the work must be re-checked before delivery. The segment grid is paged at 50 segments per page (display only); QA, the Human Gate and all counts run on the full segment set.
4. Translation memory — per client
Seeded with 3 segments only (one per pair below), each to be confirmed by a native linguist. Exact matches are 100%. Near matches show a similarity percentage with differing words highlighted.
5. Glossary / termbase — per client
Approved terms are highlighted in gold in the segment grid. If a source segment contains a glossary term but the target does not use the approved translation, QA flags it as a terminology issue — and in a regulated domain (Legal or Medical) that failure BLOCKS delivery until a human approves the exception through the Human Gate in section 7.
6. Automated QA and 7. LQA / review score
QA checks every segment for: empty target, missing final punctuation, numbers mismatch, tag mismatch, terminology, consistency (the same source translated differently), repeated words and bracket balance. The score below is a rule-based score. Draft threshold: LQA 90. Delivery is BLOCKED if project LQA is below 90 or any critical QA issue exists, until Human Gate approval.
8. Human Gate — CHECK record (delivery cannot bypass it)
Gate status (CHECK): no gate open. A gate opens automatically on a blocked QA run, or with the button below.
Approver: the pair's named roster linguist — the native linguist from the approved 2,852-pair roster (2,317 named linguists). There is no separate Quality Lead name; the roster linguist is the Human Gate approver for that pair. A gate approval never overrides a critical QA issue or an untranslated segment.
9. Chain log —
SubAgent to Agent to Supervisor to Pro Agent to Master Agent to Chief Agent to Human, logged step by step per Project with timestamps. The chain is a workflow record here.
10. Delivery and audit trail — DELIVERY ONLY
The delivery button stays disabled until a Human Gate approval is recorded, and while a Human Gate is open, while QA has not been run, or while a critical QA issue or an untranslated segment exists. A gate approval never overrides a critical issue or an untranslated segment. A delivery here is a DELIVERY.
Program view — Agile enterprise program
Program A — multi-market content rollout. One program, one quality rule, one audit trail — planned in cycles and measured every cycle. This view groups Jobs.
One quality rule for every cycle: draft threshold LQA 90; Human Gate approver: the pair's named roster linguist (from the approved 2,852-pair roster); reviewer differs from translator; TM net-rate scheme: New 100%, 95–99% 80%, 85–94% 60%, 75–84% 50%, 50–74% 30%, exact/repetition 20% translation / 100% review. A gate approval never overrides a critical QA issue or an untranslated segment.
| Cycle | Jobs in the cycle | Cycle status | Delivery rule for the cycle |
|---|---|---|---|
| Cycle 1 — Documents, first market | en to ar-SA, Legal — the Roster pair in section 2 | Planned | Blocked until this cycle's Human Gate approval AND zero critical AND zero untranslated |
| Cycle 2 — Documents, second market | en to es-ES, Marketing — the Roster pair in section 2 | Planned | Same rule — blocked until gate approval AND zero critical AND zero untranslated |
| Cycle 3 — Documents, third market | en to fr-FR, Technology — the Roster pair in section 2 | Planned | Same rule — blocked until gate approval AND zero critical AND zero untranslated |
How every cycle is measured: segments translated, LQA score against the 90 threshold, critical QA issues, untranslated segments, and the recorded Human Gate decision (decider name + reason). A cycle is measured when it closes; a failing measure sends the cycle back for rework and a fresh QA run — it never lowers the threshold.
What agility never waives: roster-before-sales — a cycle whose pair has no Roster row shows WAIT — NOT SOLD and is not assigned; the Human Gate; reviewer differs from translator; and the delivery block above. This view has no delivery button: delivery still happens only through section 10's DELIVERY button under the same rules. All program measurements are written to the same audit trail as section 10 (one audit trail for the whole program).
11. Metering and rate card
Rate card figures below are the launch card. Per-word rates are charged by the complete word — 1 credit per complete word, spaces never counted — see the rate card on the Pricing page. The discount applies to platform charges only, never to linguist pay.
| Tier | Launch | Regular (from year two) | Credits / year |
|---|---|---|---|
| Starter | $480 | $750 | 100,000 LinguaCredits |
| Growth | $9,600 | $15,000 | 1,600,000 LinguaCredits |
| Enterprise | $24,000 | $39,000 | 4,000,000 LinguaCredits |
metering rule: website fragments are counted once; editing your own translation consumes no extra credits.
11A. Software strings — /
key/value editor written from scratch for this page. Keys, source strings and targets below are original content, not taken from any other product. Tags such as {name} and %s must be preserved exactly — a missing or changed tag is flagged by QA and blocks the delivery check until a human approves it (recorded in the audit trail, same pattern as section 8).
| Key | Source string | Target — edit here | Tags in source |
|---|---|---|---|
| welcome.title | Welcome back, {name}! | {name} | |
| checkout.pay_now | Pay now — %s total | %s | |
| error.network | Network error. Please try again. | None |
Software strings shown as translated. Editing a target and re-running QA is required before the check passes.
11B. E-learning — /
course module written from scratch for this page. Structure is shown in SCORM-style form (course > lessons > segments) as a structure only. Format is confirmed per course, except for this text format. The same QA + Human Gate pattern as the main editor applies: run QA, and a blocked result needs a recorded Human Gate decision (same audit-log style as section 8) before the check passes.
Course: Welcome to Your First Lesson Set — 3 lessons, text only.
| Lesson | Source segment | Target — edit here |
|---|---|---|
| Lesson 1 — Getting started | Welcome to the course. You will learn at your own pace. | |
| Lesson 2 — Practice step | Try the practice task, then check your answers. | |
| Lesson 3 — Review step | Review what you learned and repeat any step you missed. |
Course content shown as translated. Per-course format is confirmed per course.
11C. Video subtitles — /
timed-text editor written from scratch for this page. Times are SRT-style (hours:minutes:seconds,milliseconds). The timing check flags overlapping times and out-of-order times, and empty targets. No video file is used, no dubbing is claimed, and no voice locale is claimed.
| # | Start | End | Source line | Target line — edit here |
|---|---|---|---|---|
| 1 | Welcome to this short lesson. | |||
| 2 | We will walk through three simple steps. | |||
| 3 | Pause here and try the step yourself. |
Subtitles shown as translated.
11D. Design files — /
layout-check module written from scratch for this page. Each block has a stated source-block capacity in characters. The stated threshold is: a target longer than its source block capacity is a layout overflow. An overflow blocks the layout check until a human approves it, and that approval is recorded in the same audit-log style as section 8.
| Block | Source text | Block capacity, characters | Target — edit here | Target length |
|---|---|---|---|---|
| Header block | Welcome to LinguaEmpire | 48 | — | |
| Button block | Start now | 24 | — | |
| Notice block | Your project is ready to review. | 64 | — |
Design file shown as translated. Per-file-type layout support is confirmed per file type.
12. What runs in the live platform
- MT engines run in the live platform. Scores are rule-based scores.
- The linguist network, the roster of 2,852 pairs and client jobs run in the live platform.
- Billing and payouts run through licensed payment partners in the live platform. Trial credits and billing are live; card payment receipt is not connected yet.
- The Public API + Webhooks integration is live on the platform (/api/v1). SSO and audit certification are not claimed.
- SOC 2 Type II and ISO 27001 are not certified.
- The Marketplace runs in the live platform. The 75,000+ native linguist figure is the agency network claim; the roster of 2,852 pairs is approved, with 2,317 named native linguists as the Human Gate approvers.
- Every delivery on this page is a DELIVERY, and every record is a CHECK record.
- Those items run in the live platform.
- Scale: this page stores its example data in this browser only and pages its segment table at 50 segments per page (display only — QA, the Human Gate and counts still run on the full set), so it is still NOT built for large real files yet. The platform design behind this editor and the live Lingua CAT Tool is: chunked, resumable uploads; streaming parsers that read a file piece by piece; background job queues; server-side pagination and virtualised rows; indexed TM and glossary search; object storage for files. That design targets no practical limit for real translation workloads — a literal limit-free system does not exist, and its ceilings (chunk size, job timeouts, per-plan fair-use quotas to be confirmed) are stated in the design, not hidden.
/ CHECK data. Accounts and ordering are live in the platform. Original LinguaEmpire content; no other platform's text, code, logo or design is copied on this page.
