Custom.MT

Custom.MT is a customizable machine translation platform designed for localization teams, language service providers (LSPs), product managers, and translators. It provides access to leading MT engines (including ChatGPT) and offers tooling to fine-tune models, evaluate performance, manage translation memories, clean TMX data, acquire domain-specific datasets, and deploy on-premise or in hybrid environments. The platform emphasizes workflow integration, terminology compliance, data privacy, and the ability to tailor translation pipelines to specific industries and quality requirements.

How it works

1. Connect and manage multiple MT engines, including best-in-class providers, and use the Best Engine tool to identify the most suitable model for a given language and domain.
2. Preserve or remove tags during translation to maintain formatting in software strings.
3. Integrate with your existing CAT tools and TMS via plugins and portals for Trados, Smartling, MemoQ, Shopware, and more.
4. Train and fine-tune models with terminology and style data to reduce edit distance and improve consistency.
5. Evaluate models with automated (BLEU, WER, COMET) and human evaluations across languages, domains, and linguists.
6. Deploy MT on-premise to keep confidential content within your infrastructure.
7. Maintain and clean translation memories (TMX) to keep translations useful for translators and model training.
8. Enrich terminology and ensure compliance with international standards (IFRS, SAJ, SNOMED).
9. Acquire domain-specific datasets from public sources and marketplaces (medical, legal, patent, eCommerce).
10. Leverage open-source MT options to build your own models with latest neural network toolkits.
11. Anonymize language data to remove personal identifiers while preserving linguistic content.
12. Get strategic guidance with a localization roadmap tailored to your MT maturity level.

Features and Capabilities

• Access to best-in-market MT engines, including ChatGPT, with easy evaluation and selection tools
• Best Engine tool to quickly identify the most effective model for language/subject area
• On-premise deployment to keep sensitive data offline
• Translation Memory (TMX) Cleaning to maintain usable TM data
• Terminology Compliance to enforce consistent terminology across translations
• Data Acquisition for domain-specific training data (Medical, Legal, Patent, eCommerce, etc.)
• Open-Source Machine Translation to build and customize your own models
• Language Data Anonymization to remove PII while preserving language data
• Tag preservation or cleaning options to suit software localization needs
• Terminology enrichment and glossary management
• API portals and plugins for Trados, Smartling, MemoQ, Shopware, and more
• Terminology and style training (GenAI/LLM integration) to improve consistency and quality
• Automated and human evaluation workflows for model performance
• Custom translation portals to manage workflows for different teams and clients
• Workshops and training to upskill teams in Language AI products

Who is it for

• Localization teams and managers seeking scalable MT workflows
• LSPs delivering multilingual services to clients
• Product managers aiming to localize software and UX content efficiently
• Translators needing higher quality MT-enhanced translation processes

Safety and Compliance

• Data privacy: on-premise deployment option and data anonymization features
• Terminology compliance aligned with international standards
• Transparent evaluation metrics to monitor model performance across contexts

Getting Started

• Schedule a discovery call or start a trial to explore model training, evaluation, and deployment options
• Create an account, connect MT engines, and customize workflows to your CAT/TMS tools
• Begin with a roadmap tailored to your MT maturity level and localization goals