Keops
Online tool for human evaluation of training datasets and model outputs.
View repo
Expand
your language horizons
Expand
your language horizons
We curate and enrich multilingual corpora and train domain-adaptive models
so that every culture can share its voice without language barriers.
7.5+ PB texts
datasets in 200+ languages
250+ models
machine translation
and language models
15+ tools
open-source software
20 years
in the market
End-to-end pipelines that turn dirty text
into production-ready data
End-to-end pipelines that turn dirty text
into production-ready data
Smart Datasets
High-quality datasets tailored for your needs.
- Data collection, cleaning & normalisation
- Parallel segments & docs alignment
- Insightful analysis & evaluation
- Enrichment & synthetic data generation
- Alignment with EU regulation (GDPR/EU AI Act)
Smart Models
Advanced models for efficient training.
- Fine-tune with domain data
- Audit MT/LM quality
- Deploy on-prem or in the cloud
- Work with secured development environments
- Monitor performance
Supercharge your workflow with improved data
With cutting-edge NLP techniques, multilingual advanced solutions, and comprehensive language technology tools.
Why Prompsit?
Four features that set us apart
Unique Languages
Differentiating expertise in low-resource languages including Catalan, Norwegian Nynorsk or Afrikaans and many others.
Specialised Domains
Deep domain knowledge in legal, medical, tech & engineering, and financial sectors with industry-specific terminology and compliance.
On-Premises Deployment
Secure, on-premises solutions that keep your sensitive data within your infrastructure while maintaining full control.
EU AI Act & GDPR Ready
Alignment with European AI regulations, implementing best practices for data privacy preservation and responsible AI development.
Research mindset
Following its origins as a spin-off from the Transducens research group at Universitat d'Alacant, and for a couple of decades, Prompsit Language Engineering has been contributing to research that addresses the challenges of multilingual technology.
+1800
Cites
+30
Publications
+10
R&D Projects
An expanded massive multilingual dataset for high-performance language technologies
L. Burchell et al.
Association for Computational Linguistics (ACL)
2025Do language models care about text quality? evaluating web-crawled corpora across 11 languages
R van Noord
LREC-COLING
2024Bifixer and bicleaner: two open-source tools to clean your parallel data
M. Bañon et al.
European Association for Machine Translation
2020R&D projects Prompsit is involved in
Open-source is the right way
Explore our contributions to the open-source community where we've shared our solutions, tools, and research in the field of language technology.
AltLang
Automatic language variety converter to adapt your content to local markets.
Docs
MutNMT
Educational machine translation platform to learn neural machine translation by making.
Try demo
What the press says about us
"Prompsit aporta al proyecto su amplia experiencia en la creación, limpieza y análisis de corpus multilingües abiertos a través de proyectos como ParaCrawl, MaCoCu o HPLT, su constante apuesta y contribución a proyectos de código ...
Cadena SER – Comunidad Valenciana
Machine Translation
Read more"Si queremos una auténtica soberanía digital, debemos asegurarnos de que cualquier entidad europea, sea pública o privada, pueda adaptar el modelo a sus necesidades y cumpla con nuestra normativa de protección de datos."
El Español – Disruptores
Open-Source
Full article"Prompsit is a company specialized in Natural Language Processing (NLP) and Artificial Intelligence applied to languages, with more than 15 years of experience in the combination of languages and technology."
Diario información
Business Spotlight
See storyLet's build your next AI product together
Reach out and our innovation team will get in touch within 24 hours.
Contact us