Manuel Herranz
Manuel Herranz is the founder and CEO of Pangeanic, a European language technology and artificial intelligence company specializing in multilingual AI, AI data operations (data-for-AI), machine translation workflows, anonymization, model adaptation and sovereign AI deployments. He has been involved in the creation of some European LLMs (Spanish ALIA and low-resourced Catalan Salamandra) with datasets and managing the testing, alignment, bias detection, improving the instruct and other tasks.
His work sits at the intersection of language, data, software and public infrastructure: from early machine translation systems and multilingual corpora to AI data services, model alignment, human feedback and safe deployment in regulated environments.
From multilingual technology to operational AI systems
Manuel Herranz founded Pangeanic in Valencia in 2000 as BI Europa, the subsidiary of the now extinct B.I Corporation of Japan, acting as European Director. He joined B.I Corporation after a career shaped by engineering at Gifdings & Lewis / Ford Valencia, Rolls Royce Industrial & Marine (co-generation projects), languages and the practical demands of multilingual communication.
What began as a language services company evolved into a research-driven technology organization working on machine translation, neural language technologies, AI data operations, anonymization and enterprise AI systems.
His work has followed a consistent thread: language technologies only create value when they can be deployed in real organizations, under real constraints, with reliable data, clear governance, human supervision and measurable quality. That operational view has guided Pangeanic’s transition from translation automation to multilingual AI infrastructure.
Multilingual AI
Pangeanic’s work covers machine translation, multilingual data provision to large AI Labs, language resources, model adaptation, AI evaluation and language-aware enterprise systems across European and global languages. He personally proposed four back-to-back language technology EU projects (iADAAPTA/MT-Hub, NEC TM, NTEU and MAPA) and has partnered in many others, including ELE (European Language Equality), for Europeana, and recently, Mosaic Media with several European broadcasters.
AI data operations
Manuel’s current focus includes the data layer behind AI: sourcing, cleaning, annotation, human review, evaluation, anonymization, feedback and continuous improvement.
Sovereign deployment
Pangeanic develops technologies for organizations that need control over data, infrastructure, privacy, language coverage and AI behavior in regulated or sensitive environments.
Building the data and language layer for enterprise AI
Under Manuel Herranz’s leadership, Pangeanic has moved from early statistical and neural machine translation work to a broader AI portfolio that includes data for AI, multilingual datasets, human feedback, model alignment, MT quality estimation, anonymization and private AI deployments.
| Area | Strategic role | Related Pangeanic capability |
|---|---|---|
| AI data operations | The managed service layer that turns raw, licensed or collected data into usable material for training, evaluation, adaptation and governance. | Data for AI and AI Data Operations |
| Multilingual datasets | Text, speech, audio, image, video, OCR, parallel corpora, instruction data and evaluation data for multilingual AI systems. | Datasets for AI |
| Human feedback and alignment | Expert review, preference ranking, evaluation, RLHF workflows and behavioral refinement for domain-specific and multilingual models. | Model Alignment and RLHF |
| Evaluation and QA | Benchmark design, multilingual quality assessment, model comparison, regression testing and release validation for AI systems. | Evaluation and AI QA |
| Privacy and anonymization | Secure processing of sensitive content through multilingual anonymization, data masking and controlled deployment models. | Data Masking |
| Adaptive translation | Domain-aware machine translation, terminology control, style adaptation and quality estimation for enterprise multilingual workflows. | Deep Adaptive AI Translation |
A language technology path built over two decades
Pangeanic’s development reflects a broader change in the language technology industry: from translation memories and statistical machine translation to neural systems, multilingual AI data, custom model adaptation and AI systems that require continuous evaluation.
Pangeanic founded in Valencia
Manuel Herranz founded B.I Europa as a subsidiary of B.I Corporation, rebranding as Pangeanic in 2005. During these early years, Pangeanic was a multilingual company combining language expertise with an early interest in software, automation and machine translation.
From translation automation to machine translation technology
Pangeanic expanded its work in machine translation, corpora, customization, data selection and neural language technologies, building the technical foundations for later AI data and model adaptation services.
European digital language infrastructure
Pangeanic has participated in European projects involving multilingual language technologies, anonymization, machine translation, cultural heritage access and public-sector digital infrastructure.
AI data, anonymization and model alignment
The company’s portfolio expanded into AI data services, multilingual anonymization, human feedback, model alignment, evaluation, quality estimation and private AI deployment.
Building sovereign AI systems with European values and Ethical Data
Manuel’s current work is centered on production AI systems that combine data governance, multilingual depth, human supervision, secure deployment and operational reliability.
Applied research with commercial discipline
Pangeanic’s technology direction has been shaped by years of multilingual research, public-sector deployment, European project work and production use cases. The company’s research trail connects machine translation, data selection, anonymization, evaluation, speech resources, multilingual datasets and model alignment.
Essays on language, data and artificial intelligence
Manuel Herranz writes about the industrial shift from generic AI demonstrations to operational systems: the role of data-for-AI, RL with human feedback, multilingual evaluation, small task-specific models, governance and the changing nature of language in the age of machine-mediated communication.
Language and AI
Reflections on how large language models, translation systems and multilingual AI are changing the way organizations communicate, synthesize knowledge and operate across languages.
Data as infrastructure
Analysis of the data factories behind AI systems: collection, licensing, annotation, evaluation, feedback, governance and the quiet labor that determines model quality.
Sovereign AI
Commentary on European AI infrastructure, private deployment, public-sector systems, regulatory constraints and the need for organizations to control their data and models.
Selected areas of work
These pages provide the commercial, technical and research context for Pangeanic’s current work in multilingual AI and data-centric deployment.
Data services for AI systems that need to work in the real world
Sourcing, licensing, cleaning, annotation, review, evaluation, anonymization and governance for production AI systems.
Datasets for AIMultilingual, multimodal and domain-specific datasets
Text, speech, audio, image, video, OCR, parallel corpora, instruction data and evaluation data for AI builders.
AI Data OperationsThe operating layer between data, models and deployment
Continuous data preparation, annotation, feedback, quality control and governance for AI systems.
Model AlignmentHuman feedback and RLHF for controlled model behavior
Preference data, expert review, multilingual evaluation and alignment workflows for domain-specific AI models.
Evaluation and QATesting AI systems before they reach users
Benchmark design, multilingual QA, model comparison, regression testing and release validation.
ResearchResearch and publications behind Pangeanic’s technology
Academic, European and applied research connecting multilingual data, translation, evaluation and anonymization.
Available for selected interviews, panels and executive briefings
Manuel Herranz regularly contributes to global conversations about multilingual AI, data-centric AI, AI governance, public-sector language infrastructure, the future of translation, model alignment and enterprise deployment. For media, research, institutional or partnership inquiries, please contact Pangeanic.
Topics
Multilingual AI, AI data operations, model alignment, sovereign AI, machine translation, data anonymization, evaluation and public-sector AI deployment.
Audiences
Enterprises, governments, research institutions, AI labs, language technology organizations, media and European digital infrastructure forums.
Format
Executive briefings, interviews, panels, conference talks, webinars, strategic commentary and thought leadership articles.
Discuss multilingual AI, AI data operations or sovereign deployment
Pangeanic works with enterprises, AI labs and public institutions that need reliable multilingual AI systems, governed data pipelines, human feedback, evaluation and private deployment options.