For translation agencies and localisation platform vendors
Not a translation endpoint. A translation capability that returns the file with its layout intact.
Cloud translation APIs hand you a block of text. We hand you the finished document — chart positions, headers, footers, styles and tables where they were. Every page is inspected and routed on its own, and a second model checks the result before you see it.
Coverage and programmatic access are scoped per engagement. Start with a sample evaluation.
Four things we do differently
Everything on this site grows out of these four. They are also the four things a raw translation API does not give you.
Multilingual by your book, not by a list
We do not publish a fixed language list, because a fixed list only means we stop where the list stops. Tell us the markets you serve and we configure the language pairs for them. Run a sample through us before you commit.
- Language coverage is scoped per engagement, not sold off a menu.
- Right-to-left scripts are handled at the layout stage, not bolted on.
- Multi-target delivery: one submission can be produced into several target languages in the same run.
The layout comes back unchanged
Most tools translate a .docx and hand you text with the formatting flattened. We write the translated text back into the original file: chart positions, headers and footers, styles, numbering and tables stay where they were. This part is ours — it is not a call to somebody else's document-translation endpoint.
- In-place write-back for docx, pptx, xlsx and pdf.
- Mixed PDFs — digital pages and scanned pages inside one file — are classified as their own case, because that is what real files look like.
- Page-level region rework: circle one table, choose restore, retranslate or keep. No full re-run for one bad block.
Each document gets its own processing plan
A single fixed pipeline treats page 1 and page 40 the same way. Ours does not. Every page is inspected and routed on its own: a scanned page goes down the OCR path, a chart-heavy page goes down the reflow path, a page that fails the quality check is sent back for repair before delivery.
- Page-level routing, not file-level. One document can use several paths.
- Quality checking is part of the pipeline, not an add-on.
- The visual check runs on a different model than the one that produced the translation — a translator does not grade its own paper.
- Batches resume from the last completed page after an interruption.
Five source classes, five different paths
Images, digital PDFs, scanned PDFs, mixed PDFs and text documents are not the same problem, so they do not share a pipeline. Scanned pages go through optical recognition with layout and table regions recovered; image quality is improved before recognition when needed.
- Optical recognition returns text blocks with coordinates and confidence, plus layout, table and formula regions.
- Image processing runs on GPU with a CPU fallback, so a busy queue does not stall a batch.
- A vision pass inspects the finished pages for problems that a text-only check cannot see.
Two kinds of buyer, one capability
We work with the people who deliver translation to somebody else. Which of these is you?
You run a translation agency
Your problem is capacity, turnaround and consistency across a team — not whether a model can translate a sentence. You need volume handled without your project managers becoming machine operators.
For translation agenciesYou build a localisation product
Your problem is that your customers keep asking for document translation inside your product, and stitching a raw MT API into it produces output nobody wants to ship.
For platform vendors
What we can take on
Listed as capabilities, because that is what they are. Nothing here is a self-service button — each one starts with a conversation about what you actually need.
Document translation at volume
Up to 50 files per submission and up to 1000 pages per PDF. Interrupted batches resume from the last completed page instead of starting over.
Format preservation
In-place write-back for DOCX, PPTX, XLSX and PDF. Mixed PDFs — digital and scanned pages in one file — are treated as their own case, because that is what real files look like.
Programmatic access
API and agent access, scoped to the engagement. Tell us the interface your system needs and we will tell you what it takes.
TMS and platform integration
We map our capability against your existing stack rather than asking you to replace it.
Glossary enforcement
Upload a term list in TXT, CSV or XLSX and it is applied during translation. Up to 2000 terms per glossary.
Private deployment
Assessed per project. The platform is containerised, so it can be moved — but a delivery is a project with licensing, documentation and handover, not a download.
The part most tools get wrong
A .docx is not a bag of words. It has structure, and the structure carries meaning.
Ask a text translation API to handle a document and you usually get one of two outcomes: a block of translated text with the layout gone, or a file that opens but has rearranged itself — shifted tables, broken numbering, a header that no longer fits.
We write the translated text back into the original file. The layout you sent is the layout you get back. This is the single most common reason agencies come to us after trying a raw MT API.
It is also the reason we ask for a real file rather than a description. Layout problems are visible in the output, not in a feature list.
Each page gets its own plan
One pipeline for a whole document treats page 1 and page 40 the same way. Real documents do not work like that.
Every page is inspected before it is processed. A scanned page goes down the optical recognition path. A chart-heavy page goes down the reflow path. A page that fails the quality check is sent back for repair before delivery, not after you complain.
The visual check runs on a different model from the one that produced the translation. A translator should not grade its own paper, and in practice this is what catches the errors a text-only check walks past.
One document can use several paths. That is the point.
Us, a raw translation API, and a TMS
Three different things, often confused. This table compares capability only — we do not discuss anyone's pricing, including ours.
| Row label | Us | Raw MT / cloud translation APIs | TMS platforms |
|---|---|---|---|
| Multilingual | Configured to your markets | A fixed list | A fixed list |
| Format preservation | Layout written back in place | Text or a rough file | Not an engine |
| Per-page processing plan | Each page routed on its own | One call, no quality control | Manual workflow |
| Multimodal input | Five source classes, five paths | Mostly text only | No |
| What you receive | Finished files | A block of translated text | Process management |
| Who owns the quality | Us | You | You |
What we will not claim
There is a list of things this site deliberately does not say. It is worth reading, because it tells you how to read everything else here.
We do not publish accuracy percentages, speed figures or service-level numbers, because we have not measured them under conditions that would make them honest. We do not publish a language count, because coverage is configured to your markets and a fixed number would only tell you where our list stops.
We do not publish prices. This is a capability business — what a document costs to process depends on what is in it, and quoting a number before looking at your files would be guessing.
And we do not name clients. Agencies work under confidentiality with their own customers, and we extend the same courtesy to them.
Questions we get before the first call
Can we try it ourselves?
No — there is no self-service signup, and that is deliberate. Every engagement starts with a sample evaluation: you send a real document, we run it and return the finished file. It takes one exchange and tells you more than clicking around a demo would.
Which languages do you support?
Coverage is configured to the markets you serve rather than published as a fixed list. Tell us the pairs you need and we will run a sample through them so you can judge the output.
How is this different from a cloud translation API?
Three things, in the order agencies tend to care about: the layout comes back intact, each page is routed on its own instead of the whole file getting one treatment, and the output is quality-checked by a separate model before delivery.
What file types do you accept?
PDF, DOCX, PPTX, XLSX, DOC, PPT, XLS, JPG and PNG, up to 500 MB per file. Scanned documents and image-only pages go down the optical recognition path.
Do you replace our CAT tool?
No. Most engagements run alongside the tools you already have. We map our capability against your stack rather than asking you to move off it.
Can it be deployed inside our own infrastructure?
It can be assessed. The platform is containerised, which makes a private deployment technically possible, but a delivery involves licensing, deployment documentation, handover and offline model arrangements. It is a project, not a download.
Send us a real file. See what comes back.
Pick the document that normally causes you the most trouble — the scanned one, the chart-heavy one, the one with the tables. We will run it and send the finished file back so you can judge the layout yourself.
- No signup, no self-service portal — you send a file, we run it.
- You get the finished document back, not a screenshot of one.
- If it is not good enough, you have lost nothing.