Experts in
Language Engineering

NLP & AI at your service

Through language processing, we turn language data into services: automatic translation, text classification, text mining, NER, anonymisation, etc.

Language technology solutions

We help you to improve translation and localisation processes, solve linguistic doubts, teach language technology, structure unstructured data, perform linguistic evaluation and much more.

High-quality datasets

We gather and curate high-quality multilingual datasets: we compile, process, analyse, clean, anonymise and evaluate them for you. In any format, in any language.

Our products and services

KEOPS

Manage datasets and models evaluation campaigns

AltLang

Adapt your content to different language variants

MutNMT

Learn the basics of NMT systems

TermGallery

A place for words and their usage

Corset

Search through datasets, create custom datasets

Bifixer

Fix datasets and detect near-duplicates

Bicleaner

Clean and score parallel corpora

Biroamer

Anonymize and mix your datasets

Bitextor

Automatically harvest bitexts from multilingual websites

Monotextor

Build monolingual datasets from websites

Opus Trainer

A machine translation/language model data trainer

Opus Cleaner

Semi-supervised dataset cleaner

Apertium

A free/open-source machine translation platform

HPLT Analytics

Get insights on your datasets content

Partners in research and innovation projects

What defines us

Language diversity

We care about languages, no matter the size or shape of their resources. We are at ease working with different alphabets, morphology, lemmas, grammar structures, glossaries, concordances, corpora, domains, etc. We make the most of any piece of text in any language, specially in multilingual scenarios.

~2.5PB+ datasets

for around 100 languages, targeting low-resourced ones

Research DNA

Our roots are in the Transducens research group of the University of Alacant. A mixed team of computer engineers and linguists started Prompsit back in 2006. Since then, we’ve been serving market interests transforming research ideas into final products, by performing applied engineering.

250+ models

including machine translation and language models

Open-source & value-added

If others did it and shared it, we will keep doing and sharing! This way, we add value to what already exists. As open-source contributors for decades, we do not reinvent wheels, we shape and ensemble them to effectively deliver value-added final products.