The work,
in numbers.

A returning Language Solutions client came to us with a super-urgent production job: 394 separate files totalling 1,626 pages, all needed for a live translation project on a timeline measured in days, not weeks. Volume at this scale usually gets broken up and handed across multiple vendors — which is exactly where consistency, formatting and version control start to slip. The client needed the whole set back clean, complete and on time.

We kept the batch together under one team and one file register rather than splitting it across uncoordinated operators. Work ran across our three shifts so the file set moved continuously through prep, formatting and QC. A single project owner tracked all 394 files against one checklist and stayed on email with the client throughout, so status was visible at the file level, not just at milestones. Every file was verified before release — no partial handovers, no rework loop.

An EN-US to ZH-CN OCR review job arrived as a 434-page Clinical Trial Agreement bundle — fifteen separate document parts including medical records, hospitalisation reports, and expert assessments, several with French source material. The client's translation pipeline needed clean, structured English text before the Chinese conversion could begin. The catch: ABBYY FineReader could not extract the content cleanly. Pages were blurry. Tables were split across cells. Special characters came through as junk. Some pages were partly handwritten. Both portrait and landscape orientations sat inside the same file. The original deadline gave us 48 hours.

First, we said no — honestly. The original timeline wasn't deliverable at the quality the client needed, and splitting the work across uncoordinated operators would have broken table continuity. We documented the file complexity, negotiated a workable deadline, and committed to it. Then we ran the operation: tried multiple OCR engines on each problem page (ABBYY and alternatives), and on every page where OCR fell short, switched to manual key-and-verify, line by line. Operators zoomed in to read minute text. Split tables were reconstructed cell by cell. Handwritten passages were transcribed by hand. The project manager stayed on email with the client every day — not just on milestones, but on the daily state of the file.

A major academic publisher needed high-volume STM journal composition — including complex MathML equations, chemistry structures, and multi-language scripts — at a 48-hour standard TAT, with simultaneous JATS XML and print PDF output, and WCAG AA EPUB 3 production for digital distribution.

Dedicated compositor team with specialist STM training. XML-first workflow producing simultaneous JATS XML and print-ready PDF from a single source. EPUB 3 with MathML, accessible alt text, and EPUBCheck 5.1.0 validation. IDML-based AI pre-check running on every file before operator involvement.

A European LSP won a contract to localise a pharmaceutical patient information leaflet into 22 languages — including Arabic, Hebrew, Japanese, Chinese (Simplified and Traditional), and 17 European languages. Previous attempts at Arabic and Hebrew had resulted in font failures, kashida handling errors, and bidirectional text issues that required full rework.

B2K ran font compatibility checks across all 22 languages before translation began, flagging font gaps and recommending replacements. XLIFF round-trip with TransTools tag cleaning for all 22 files. RTL specialists handled Arabic and Hebrew layouts including kashida, tatweel, and bidirectional text. Layout space prediction ran on all language pairs before translation to pre-solve text expansion issues.