Bergamot · Project

Private Machine Translation Running in Your Browser — No Cloud, No Data Leaks

digitalPilotedTRL 7

Imagine Google Translate, but it runs entirely on your own computer instead of sending your text to someone else's server. The Bergamot team built translation technology that works right inside your web browser — like Firefox — so nothing you type ever leaves your machine. It can even help you fill out forms in a language you don't speak, and it tells you how confident it is in each translation. Think of it as a personal interpreter sitting inside your laptop, working offline and keeping all your secrets.

By the numbers

consortium partners across Europe

countries in the development consortium

project deliverables completed

working demo user interfaces delivered

industry partners in the consortium

The business problem

What needed solving

Organizations handling sensitive documents in multiple languages face a painful choice: use cloud translation services and risk exposing confidential data, or pay for expensive human translators for every document. Industries like legal, healthcare, and government cannot afford data leaks, yet operating monolingually in a multilingual European market means lost business, slower processes, and frustrated customers.

The solution

What was built

The team built a fully functional browser-based machine translation system that runs on local hardware without any cloud connection. Key deliverables include 3 demo user interfaces — one for general translation, one with built-in quality estimation so users know when to trust the output, and one for outbound translation enabling users to write in languages they don't speak. The system includes domain adaptation that customizes translations to specific websites or content types on the fly.

Audience

Who needs this

Law firms handling cross-border litigation and multilingual contractsGovernment agencies processing citizen requests in multiple EU languagesHospitals and health insurers managing cross-border patient recordsDefence and intelligence organizations requiring air-gapped translationFinancial compliance teams reviewing foreign-language regulatory documents

Business applications

Who can put this to work

Legal & Professional Services

any

Target: Law firms and consultancies handling cross-border contracts

If you are a law firm dealing with confidential multilingual contracts and client documents — this project developed browser-based translation that never sends your text to external servers. Your lawyers can translate foreign-language evidence, contracts, and correspondence without risking client confidentiality. The built-in quality estimation tells users how reliable each translation is before they act on it.

Public Administration & Government

enterprise

Target: Government agencies and municipalities processing multilingual citizen requests

If you are a government agency dealing with citizens who submit forms and requests in multiple languages — this project developed a free, open-source translation tool integrated with web browsers that runs entirely on local machines. With 3 dedicated user interface deliverables including outbound translation, civil servants can respond to foreign-language inquiries without procuring expensive cloud translation services or violating data protection rules.

Healthcare & Insurance

mid-size

Target: Hospitals, clinics, and health insurers processing patient records across borders

If you are a hospital or health insurer dealing with cross-border patient records containing sensitive medical data — this project developed client-side translation with domain adaptation that adjusts to specialized medical terminology on the fly. Patient data stays on your local machine instead of being uploaded to cloud translation services, helping you comply with strict health data regulations while serving multilingual patients.

Frequently asked

Quick answers

How much does this translation technology cost to deploy?

Bergamot was developed as free, open-source software designed for integration with open-source browsers like Mozilla Firefox. There are no licensing fees for the core technology. Deployment costs would mainly involve IT setup and any customization for your specific domain.

Can this handle the volume of translations a large organization needs?

The technology is designed to run on standard desktop hardware with compact model downloads, meaning each employee's machine handles its own translations independently. This scales naturally — adding more users doesn't overload a central server because there is no central server. Based on available project data, the efficiency research specifically targeted running neural machine translation on consumer-grade hardware.

What is the intellectual property situation — can we use this commercially?

Bergamot was funded as a Research and Innovation Action (RIA) and produced free, open-source software. The project explicitly aimed for bottom-up adoption by non-experts. Specific licensing terms should be confirmed with the University of Edinburgh as coordinator, but the open-source nature suggests permissive commercial use.

How do we know the translations are accurate enough for business use?

The project developed dedicated quality estimation technology — one of 3 demo deliverables is specifically a user interface with quality estimation built in. This means the system flags uncertain translations rather than presenting everything with equal confidence, letting your team focus human review where it matters most.

Does this comply with GDPR and data protection regulations?

This is where Bergamot stands out from cloud alternatives. Because all translation happens locally on the user's machine, no text is ever transmitted to external servers. The project objective explicitly states this design preserves citizen privacy and addresses sectors that require confidentiality. No data leaves your network.

How long would it take to integrate this into our existing systems?

The technology was built as a browser extension for Firefox with 3 working user interface deliverables completed by project end in June 2022. For browser-based use, deployment can be as simple as installing an extension. Custom integration into internal web applications would require additional development work depending on your setup.

Does it support domain-specific terminology in our industry?

Yes — the project included dedicated research on dynamic domain adaptation. Based on the project objective, the system adapts translation on the fly using local context from the website or user, meaning it learns the specific writing style and terminology of your documents without uploading that proprietary information to the cloud.

Consortium

Who built it

The Bergamot consortium brings together 6 partners from 5 countries (CZ, DE, DK, EE, UK), led by the University of Edinburgh — one of Europe's top computational linguistics research centres. With 4 universities and 2 industry partners (33% industry ratio), the consortium is research-heavy but the industry involvement signals real-world testing and deployment expertise. The cross-European spread across Czech Republic, Germany, Denmark, Estonia, and the UK reflects diverse language market needs. For a business buyer, this means the technology was developed with multiple European languages and real deployment constraints in mind, not just as an academic exercise.

THE UNIVERSITY OF EDINBURGHCoordinator · UK
TARTU ULIKOOLparticipant · EE
MZ DENMARK APSparticipant · DK
THE UNIVERSITY OF SHEFFIELDparticipant · UK
MZ DENMARK GMBHparticipant · DE
UNIVERZITA KARLOVAparticipant · CZ

How to reach the team

The project coordinator is The University of Edinburgh (UK). SciTransfer can facilitate a direct introduction to the research team.

Order a report on this consortium See THE UNIVERSITY OF EDINBURGH profile →

Next steps

Talk to the team behind this work.

Want to explore how private browser-based translation could work in your organization? SciTransfer can arrange a technical briefing with the Bergamot team and help assess fit for your specific use case.

Order a report on this topic More in Digital & ICT