Claude · DeepL · OpenAI · Gemini

Fill missing translations. Import the same day.

Upload your .sdlxliff or .mqxlz. AI fills the gaps with your glossary enforced and TM matches untouched. The result imports cleanly. No tag repair needed.

No credit card required

Works with:SDL Trados.sdlxliffmemoQ.mqxlz
contract-2024.sdlxliff3 translated
#Source (EN)Target (DE)Status
1This Agreement is governed by the laws of Germany.Diese Vereinbarung unterliegt dem Recht Deutschlands.confirmed
2The parties agree to resolve disputes by arbitration.Die Parteien einigen sich auf Schiedsverfahren.translated
3All warranties are expressly excluded.Alle Gewährleistungen sind ausdrücklich ausgeschlossen.translated
4Liability is limited to direct damages only.translating…
4 of 48 segmentsTranslating…
2
CAT formats
4
AI providers
100+
Language pairs
50%
Batch savings

Why generic AI translation breaks CAT workflows

Three failures that happen when tools weren't built for XLIFF.

01

Tags break on import

Inline XLIFF tags get treated as text. One misplaced tag corrupts the whole file. You fix it manually or start over.

02

Glossaries get ignored

AI picks statistically likely words over your approved terms. Domain terminology built over years comes back wrong.

03

TM matches get retranslated

High-match segments get sent to the provider, charged for, and often returned as something different.

These aren't edge cases. They're why translators stop using AI tools after the first trial.

Upload. Translate. Reimport.

Three steps. LLM Catalyst fits between your CAT tool export and import.

01

Upload your XLIFF

Export from SDL Trados or memoQ and drop it in. Segments, TM scores, and metadata parse automatically. No format conversion. No copy-paste.

File upload

contract-2024.sdlxliff

SDL Trados · 2.4 MB

Claude
0 / 3 translated

The parties agree to resolve disputes by arbitration.

Translating…

All warranties are expressly excluded.

Liability is limited to direct damages only.

02

AI fills the gaps

Pick Claude, DeepL, GPT, or Gemini. Your glossary terms inject at inference. TM matches at or above your threshold stay locked and unchanged.

03

Review and reimport

QA runs automatically before download. Issues surface in the dashboard. Edit inline if needed, then download and reimport. Tags are intact.

QA · 36 segments
Tag integrity
Glossary compliance
Length ratio
Untranslated content
contract-2024.sdlxliff

Built around the constraints that matter

Every feature exists because a CAT workflow requires it.

QA runs before you download

Four checks on every segment. Issues surface in the dashboard, not after reimport.

QA Report0 / 4
Tag integrity
Glossary compliance
Length ratio
Untranslated content
Checking segments…

Tags survive. Every time.

Tags become placeholders before AI sees the text. A repair pass puts them back.

SourceHello <bpt>world</bpt>
AI seesHello {T:0}world{T:1}
OutputHola <bpt>mundo</bpt>

Glossary at inference

TBX or CSV terms inject into every call. Violations flagged before download.

TM threshold

Set per project, override per document. Matched segments stay locked.

Four providers

Claude, DeepL, GPT, or Gemini. Batch mode on Claude and OpenAI cuts costs 50%.

Four AI providers, one workflow

Switch per translation run. No reconfiguration.

Claude

Anthropic

Batch mode

DeepL

Bring your own key

OpenAI GPT

Batch mode

Gemini

Google

Transparent pricing

Free to start. Upgrade when your volume grows.

Free

  • 5,000 segments/month
  • 1 active project
  • All 4 AI providers
  • Full QA suite
Most Popular

Pro

  • 50,000 segments/month
  • Unlimited projects
  • Batch mode (50% savings)
  • Glossary management
  • Your own DeepL key

Team

  • 200,000 segments/month
  • 5 seats
  • Team roles
  • TM export
  • Priority support
See full pricing

Questions translators actually ask

Especially about tags, TM matching, and what happens to client files.

Try it on a real file.

Upload a .sdlxliff or .mqxlz. See the translated output before you commit. Free, no card required.