Reginald is a web-based computer-assisted translation (CAT) tool for professional translators — it splits documents into segments, pairs each source with its translation, and keeps your work consistent with translation memory and terminology.Reginald is a web-based CAT (Computer-Assisted Translation) tool. You upload documents (DOCX, XLIFF, MXLIFF, MQXLIFF, SDLXLIFF), it splits them into segments, and you translate segment by segment. It stores your translations in a Translation Memory database and suggests them when similar text appears later. It also connects to machine translation APIs (Google, OpenAI) for initial suggestions. A terminology panel shows relevant terms. When done, you export back to the original format. The interface has three columns: source, target, terminology. Keyboard navigation works throughout.
What Reginald Is
Reginald is a CAT tool that runs in a browser. It segments your documents, pairs each source segment with its target, and supports the work with memory, terminology, and machine translation.Reginald is a CAT tool that runs in a web browser. CAT stands for Computer-Assisted Translation—software that helps translators work with text by breaking documents into segments, storing past translations, and managing terminology.
File Handling
You upload files through the browser. Reginald extracts the translatable text, segments it, and lets you download the finished translation in its original format.You upload files through the browser. Supported formats include Microsoft Word (.docx), XLIFF, and bilingual CAT formats. Reginald parses them, preserves tags and formatting, and presents segments in the editor. When you finish, it reconstructs the file in the original format for delivery.
Translation Memory
Every confirmed translation goes into a Translation Memory (TM). When you work on new text, Reginald searches the TM for matches. Exact matches (100%) appear first, then fuzzy matches with percentage scores. You can import existing TMX files from other tools and export your TM as TMX.
Machine Translation
Reginald connects to external machine-translation providers (OpenAI, DeepL, Google). MT is a starting point you confirm or edit — never an automatic replacement for your judgement.Reginald connects to external MT providers: Google Translate (NMT and their newer LLM-based translation), OpenAI GPT-4o, and an adaptive MT service. When you open a segment, it fetches MT suggestions if no good TM match exists.
terminology
There is a terminology database (termbase) that flags approved terms as you translate, helping you stay consistent with client and project vocabulary.There's a terminology database (termbase) that stores approved terms by language pair, project, or client. When you translate, matching terms appear in the right column so you can stay consistent. You can import/export TBX files or manage terms directly in the UI.
The Interface
Three columns: source text on the left, your translation in the middle, and a right-hand rail of translation-memory matches, terminology, and actions.Three columns: source text on the left, target (your translation) in the middle, terminology and metadata on the right. Navigation is keyboard-first, with shortcuts to jump between segments, confirm translations, and open concordance searches.
Multi-User Setup
Reginald is built as a multi-tenant platform: each organisation's projects, memories, and termbases are isolated, with role-based access for translators, reviewers, and project managers.Reginald is built as a multi-tenant system. Multiple users can have accounts, organized under "tenants" (organizations). Users can work on the same document simultaneously—there's locking to prevent two people from editing the same segment at once, and changes sync in real-time.
Current State
The translation-memory and terminology systems are in place, machine-translation providers are integrated, and the editor supports the day-to-day translation workflow.The TM and terminology systems work. MT integration works. Multi-user features are implemented. The system is in active development, with more automation and workflow improvements coming next.