Translation workspace · AI and human

everything, checked and approved.

Send a spreadsheet, a website, a Trados package or a million-word JSON file. Choose AI, linguists or both, set the rules once, and get every file back with the checks and sign-offs recorded.

Get a quote See plans
Running a translation team or agency? Book a demo
  • Plans from £25 a month
  • Managed jobs quoted per project
  • Privacy Mode for sensitive content
product-catalogue.xlsx 1,240 rows · EN → FR, DE FRDE
  • Source: column C
  • French + German
  • Keep product names in English
  • Max length: column F
  • ACME glossary
  • AI draft → AI check → human approval
#Source (EN)FrenchLengthCheck
14 Trailblazer 40L hiking backpack with rain cover Sac à dos de randonnée Trailblazer 40 L avec housse de pluie 60/65 Term kept
15 Waterproof zip and padded laptop sleeve, 1.2 kg Zip étanche et pochette ordinateur rembourrée, 1,2 kg 53/65 1.2 → 1,2
16 Summit Pro trekking poles, carbon, adjustable 105–135 cm Bâtons de trekking Summit Pro, carbone, réglables de 105 à 135 cm 65/65 At limit
17 Insulated jacket that packs into its own pocket Veste isolante légère qui se range dans sa propre poche intérieure Veste isolante qui se range dans sa propre poche 66/65 48/65 Over limit Shortened
4 of 4 approved by the French reviewer Export product-catalogue_fr.xlsx
Example project. Rows 14–17 of a French catalogue job.
Start here

Which of these are you?

  1. I need something translated

    Send the files and a short brief. Our team runs the work, has it reviewed, and returns finished files in the original format.

    Get a quote

  2. We translate every week

    Run your own projects with memories, terminology, QA and reviewers in one workspace. Plans start at £25 a month.

    Plans and pricing

  3. We run a translation business

    A CAT tool and TMS that opens Trados packages and XLIFF, runs LQA, and tracks clients, vendors and jobs.

    Book a demo

  4. We're building with it

    Start projects, upload files and fetch approved translations from your own code, CI pipeline or AI agent.

    API, CLI and MCP
How it works

Every project follows the same four steps

What changes from job to job is who does each step and how strict the checks are. You decide that per project, per client or per content type.

  1. Bring the content

    Upload files or connect Phrase, Lokalise, WordPress or GitHub. The text is extracted and the layout, tags and structure are kept.

  2. Set the rules

    Target languages, terminology, style guide, character limits, and anything that must stay untranslated.

  3. Translate and check

    AI, linguists or both. Rule-based QA runs on every segment, and scored LQA runs where you need it.

  4. Approve and deliver

    Reviewers sign off in the editor. You get the original format back, and the approved work goes into your memory and termbase.

Three common set-ups

AI-led. Fast and cost-controlled, with people where they add most.

  1. ExtractAutomatic
  2. Apply memory and terminologyAutomatic
  3. Draft translationAI
  4. QA and scored LQAAI
  5. Review flagged segmentsLinguist
  6. DeliverAutomatic

Linguist-led. For regulated, creative or high-visibility content.

  1. ExtractAutomatic
  2. Apply memory and terminologyAutomatic
  3. Translate with AI suggestionsLinguist
  4. ReviseReviewer
  5. QAAutomatic
  6. Client approvalClient

Fully automated. For high-volume, low-risk content, with a quality threshold instead of a person.

  1. Sync from connectorAutomatic
  2. Draft translationAI
  3. QA and scored LQAAI
  4. Fix segments below thresholdAI
  5. Spot-check sampleLinguist
  6. Push back to sourceAutomatic
In the product

The difficult jobs, handled in the same workspace

The memory and the glossary sit beside every sentence

Open a line and the closest match from your translation memory is already there, with the differences marked. Product names from the termbase light up in the source, so nobody translates the brand.

  • Fuzzy matches shown with what changed
  • Termbase entries highlighted as you work
  • Approve with Ctrl+Enter and move to the next line
The translation editor

Every file, language and stage on one page

Each project shows its files by language, the stage each one is at, who has it and how far they have got. Progress is counted from approved lines, so the number means something.

  • Files and languages side by side
  • Stages, assignees and deadlines
  • Memories, termbases and QA settings for the project
How teams work together

The transcript, the subtitles and the voice on one timeline

Upload a video and LanguageOps transcribes it into timed subtitles, labelled by speaker. Translate them with your glossary, adjust the wording and timing side by side, and add a dubbed voice where the audience would rather listen than read.

  • Transcription with speaker detection
  • Subtitles edited against the picture
  • Export SRT and VTT, or dub block by block
Video localisation

Take the SDLPPX. Send back the package.

Clients and agencies send Trados packages. LanguageOps reads the project file, finds the target bilinguals, and rebuilds the original package with the completed translations inside.

  • Language direction read from SDLPROJ
  • Tags protected and checked before delivery
  • Source files, reports and project settings preserved
File formats and packages

A score you can show the client

Rule-based QA catches numbers, tags, terminology and spelling on every segment without using AI tokens. Scored LQA grades the translation by category, against the threshold you set.

  • Errors by category and severity
  • Pass or fail against your threshold
  • Reviewer comments attached to the segment
Consistency, QA and LQA

LQA · French · 4,812 words

97.2/100

Pass · threshold 95
CategoryMinorMajorCritical
Accuracy200
Terminology110
Fluency300
Style100
Locale conventions000
Segment 212 · Terminology · Major

tableau de bord console d’administration: the client glossary uses “console d’administration” for “admin dashboard”.

Files that crash desktop CAT tools

Open a 650,000-word JSON export or a 100,000-row spreadsheet. Choose which keys or columns to translate, split the work, and load segments progressively in the editor.

  • Filter keys and columns before you pay for words
  • Split one file across several linguists
  • Segments load progressively, so the editor stays fast
Large-file translation

catalogue-strings.json · 41 MB · 650,112 words

products.*.titletranslate
products.*.descriptiontranslate
products.*.skuskip
products.*.image_urlskip
meta.*.updated_atskip

247,040 words to translate after filtering, split into 8 parts

  1. Part 1100%
  2. Part 2100%
  3. Part 3100%
  4. Part 486%
  5. Part 564%
  6. Part 641%
  7. Part 712%
  8. Part 80%
Beyond documents

Video, speech and search are translation too

The same memories, glossaries and reviewers work on a film, a live meeting or a page you want found in another market.

Video and audio

Subtitle, dub and interpret in the workspace your documents already use, so a product name is translated the same way on screen as in the manual.

Content and search

Find what readers in each market are searching for, write or transcreate for it, and see how your pages show up in search results and AI answers.

Used for technical and creative translation by teams at

  • Barclays
  • Primark
  • Converse
  • UK Department for International Trade
  • Plykea
  • Saatchi & Saatchi
We tested our most complex workflows through LanguageOps, and they all passed. Rare file formats, multiple column spreadsheets to LaTeX files, custom character limits - the system handled it all. The automated QA and LQA features really improve efficiency on large scale jobs.
Lisa Willliams, CEO, ALM Translations
What's in the workspace

The tools you'd otherwise buy separately

Every paid plan includes all of it. Heavy AI use is metered, so what you pay follows the work you do.

Translate and edit

Quality

Language assets

Operations

Connect

Works with the systems you already have

Pull content from your TMS, CMS or repository and push approved translations back without copying and pasting. Developers can drive the same workflow from code.

  • Phrase
  • Lokalise
  • WordPress
  • GitHub
  • Trados Studio
  • memoQ
  • Smartling
  • All connectors
REST API
Projects, files, translation, QA and exports.
CLI
Scriptable jobs for local files and CI.
MCP
Let AI agents such as Claude start and track translation jobs. Access on request.
terminallangops CLI
# Create a project and send a file for AI draft + review
langops projects create --name "Q3 catalogue" --source en --target fr,de
langops files upload product-catalogue.xlsx
langops translate run --workflow ai-draft-human-review
langops jobs watch

✓ fr  1,240/1,240 segments · QA passed · awaiting review
✓ de  1,240/1,240 segments · QA passed · awaiting review
Control

Your content stays under your control

  • Privacy Mode

    Pre-translate on a local server with your approved memories and terminology, for content that shouldn't go to an external AI provider.

  • Not used for training

    We don't train AI models on your content. Your memories, terminology and approved translations improve your own projects and nobody else's.

  • People sign off

    Approval is a recorded step. Every edit is attributed to a person and a stage, and roles decide who can edit, review and export.

  • A signed DPA

    A data processing agreement is available on request, and we'll tell you which providers a workflow uses before any content goes through it.

    Read the DPA
Pricing

Pay for the way you work

Managed job

Quoted

Send us the files. We scope, translate, review and deliver.

Get a quote

Subscription

From £25/mo

Run your own projects with every tool included. Annual plans get two months free.

Compare plans

Enterprise

Custom

Custom volume, SSO, SLA, reseller accounts and deployment options.

Talk to us

Questions people ask first

Can I use AI and human translators in the same project?

Yes. Each project has its own set-up: AI drafts with linguist review, linguists working with AI suggestions, or fully automated with a quality threshold. You can change it per client or per content type.

Will my files come back in the same format?

Yes. Documents, spreadsheets, XLIFF, JSON, PO files and Trados packages are rebuilt in their original format, with tags and structure preserved. Missing or broken tags are flagged before export.

How does Privacy Mode work?

Privacy Mode runs pre-translation on a local server using your approved memories, terminology, style guides and reference files, so sensitive content isn't sent to an external AI provider.

Do I have to buy a subscription to get something translated?

No. Managed jobs are quoted per project. Send the files and a brief, and we'll return a scope and price before any work starts.

Can we move from Trados, memoQ or Phrase?

Yes. Import TMX memories and TBX termbases, open SDLPPX packages and XLIFF directly, and connect Phrase or Lokalise to keep work flowing during the switch.

Send us a real file and watch what happens to it

Get a price for the job, or book 30 minutes and see your own file go through extraction, translation, checks and export.

Get a quote Book a demo