Custom.MT

Custom.MT is a comprehensive AI translation management platform that empowers localization teams and Language Service Providers (LSPs) to harness the full potential of modern machine translation and Generative AI. Founded in 2020, the company helps businesses implement advanced AI strategies into their localization workflows. The flagship product, Custom.MT Console, serves as a central hub to manage, evaluate, and deploy a wide range of stock and custom-trained translation models from leading providers like OpenAI, Anthropic, Google, Microsoft, and more.

The platform is designed to move beyond the limitations of standard, one-size-fits-all MT. It provides the tools to identify the most accurate translation engine for any given language pair and subject matter, and then further enhance its performance through custom training. This allows organizations to create MT models that are finely tuned to their specific terminology, style, and brand voice, resulting in significantly reduced post-editing effort and higher quality output.

How to use Custom.MT

Using Custom.MT involves a strategic approach to integrating AI into your translation process:

1. Connect Your Tools: Start by integrating the Custom.MT platform with your existing localization ecosystem. It offers seamless plugins for popular CAT tools and TMS like Trados, memoQ, Smartling, and XTM.
2. Evaluate and Select: Use the 'Best Engine' tool to upload a sample of your content. The platform automatically evaluates various MT models and provides a data-driven recommendation on which one performs best for your specific language and domain.
3. Customize and Train: To achieve maximum accuracy, provide your own data, such as translation memories (TMX) and glossaries. Custom.MT uses this data to fine-tune the selected model, teaching it your specific terminology and stylistic nuances. The platform also offers services for TMX cleaning and data acquisition if needed.
4. Deploy and Translate: Plug the winning, customized model directly into your workflow via the integrated plugins. You can choose between a secure cloud deployment or an on-premise solution for highly confidential content.
5. Manage and Optimize: Continuously manage your models through the Custom.MT Console. Monitor performance, ensure terminology compliance, and retrain models as new data becomes available to maintain peak performance.

Core Features of Custom.MT

• Centralized MT Management: A single console to manage and connect multiple MT and GenAI models from providers like Google AutoML, Microsoft Translator, OpenAI, and ModernMT.
• Best Engine Finder: An analytical tool to objectively measure and compare the performance of different MT engines for your specific content.
• Custom Model Training: Fine-tune MT models with your own translation memories and glossaries to improve accuracy, terminology adherence, and style.
• Seamless CAT/TMS Integration: Ready-to-use plugins for Trados, memoQ, Smartling, XTM, and other leading localization tools.
• On-Premise Deployment: Option to deploy machine translation on your own hardware, ensuring that confidential data never leaves your secure environment.
• Translation Memory (TMX) Cleaning: Services to clean and prepare your TMX data for optimal model training or for continued use by human translators.
• Terminology Compliance: Enrich glossaries and ensure that the MT output adheres to industry-specific or company-specific terminology standards.
• Data Acquisition & Anonymization: Assistance in acquiring in-domain training data and automatically removing personal information (PII) to comply with privacy regulations like GDPR.

Use Cases for Custom.MT

Custom.MT is versatile and addresses the needs of various sectors:

• Language Service Providers (LSPs): LSPs can create, brand, and monetize their own custom MT engines, offering superior value to clients and increasing their profit margins.
• Global Enterprises: Companies can ensure consistent brand voice and terminology across all multilingual communications, from marketing materials to technical documentation.
• Software & Tech: Efficiently translate software strings, user interfaces, and help-desks at scale, with advanced tag processing to preserve code integrity.
• Healthcare & Life Sciences: Train models on specialized medical data (e.g., SNOMED) for highly accurate and compliant translations of clinical trial documents, patient information, and healthcare apps.
• Legal & Finance: Securely translate confidential documents using on-premise deployment and models trained on legal or financial (e.g., IFRS) terminology.
• Gaming: Provide real-time, in-game chat translation and localize vast amounts of game content quickly and cost-effectively.

Advantages of Custom.MT

The primary advantage of Custom.MT is the shift from using generic MT to controlling a customized, high-performance translation solution. Key benefits include:

• Superior Quality: Custom-trained models deliver more accurate and stylistically appropriate translations.
• Increased Efficiency: Drastically reduces the time and cost associated with manual post-editing.
• Enhanced Security: On-premise deployment options provide maximum control over sensitive data.
• Flexibility: A single plugin provides access to a multitude of MT providers, simplifying multilingual project setup.
• Positive ROI: Clients report achieving a positive return on investment within 3-4 months by increasing margins and throughput.
• Competitive Edge: LSPs and enterprises can differentiate themselves by offering highly specialized and efficient translation services.

Pricing and Plans

Custom.MT offers a 'Start Free Trial' option to allow users to experience the platform's capabilities. As the services are highly tailored to individual needs, specific pricing is provided on a custom basis. Potential clients are encouraged to 'Schedule a discovery call' or 'Book a demo' to discuss their requirements and receive a personalized quote. Plans are designed to scale with your business, considering factors like translation volume, number of users, and the need for advanced services such as model training, on-premise deployment, or data acquisition.