Meet Custom.MT at LocWorld56

Bring your own dataset, pick the engines to compare and get a side-by-side benchmark. Make language quality measurable and prove your choice to leadership.

• Every engine side by side on your data
• Errors flagged in context: accuracy, terminology, fluency, formatting
• Segment-by-segment review across engines
• Quality tracked over time for every language pair

Set up each workflow step by step: choose the engine for the job, protect tags, apply your glossary and add AI editing. Or just ask the AI Assistant to design the workflow in plain language.

• MT or LLM engine chosen per workflow and language pair
• Tag and placeholder protection
• Your terminology applied to every translation
• Style rules written once in plain language
• Works inside Trados, memoQ, Smartcat, XTM, Phrase and Smartling

Create a terminology project, set the domain and subdomain, and get extracted terms in minutes. Review, approve and export a clean termbase.

• Terms extracted automatically
• Domain-aware definitions and translations, ready to review
• Approve or reject every term; no duplicates in your glossary
• Export approved terms to CSV, XLSX or TBX

Generic benchmarks don't show how a QE model behaves on your content and domain. Rigel turns your translation memory into a realistic test set and measures each model on it.

• QE models compared on your content, not generic test sets
• Risk measured: false positives and missed errors
• Suggested thresholds: balanced, economical or safety-first
• Re-run over time as prompts, models and content change

A standalone, CAT-agnostic service. Send your TMX exports and get back a dataset ready for MT training, quality estimation and CAT workflows.

• 40+ quality filters, including neural ones
• 70–80% of incorrect segments identified in a routine scan
• Duplicates, encoding issues, Unicode noise and glued words removed
• Your high-quality segments kept intact