@Pangeanic

European multilingual AI company building the trust infrastructure between enterprise knowledge and artificial intelligence through AI Data Operations + AI QA

 Valencia, España
pangeanic.com
Joined August 2009
  • Tweets4.3K
  • Following346
  • Followers1.4K
521Media
ISO 18587 is starting to catch up with AI. The draft now refers to “non-human translation output” The real divide is no longer human vs. machine. It is controlled vs. uncontrolled production. pangeanic.com/blog/iso-18587… #ISO18587 #AITranslation #MTQE

ISO 18587 Draft: AI Translation, LLMs & Human Review | Pangeanic

ISO/DIS 18587 expands post-editing to non-human translation. See what it means for LLM translation, MTQE, risk-based human review, enterprise AI workflows

pangeanic.com
1
15
Quality filters that work in English degrade measurably in every other language you cover. Tested across 35 languages. The vendor question is not how many languages they cover. It is how quality is measured per language pair. pangeanic.com/data-for-ai/ev…

Evaluating Multilingual AI Data Services | Pangeanic

How to evaluate multilingual AI data services: annotation quality per language, data governance, privacy architecture, and sovereign deployment.

pangeanic.com
1
2
63
Which is the most translated #website in the world? pangeanic.com/knowledge_cent… #translationtuesday #translatingeurope #translatebutton – at Spain

What is the most translated website in the world? | Pangeanic

If you ever wondered what the most translated website in the world is, here's the winner you did not expect: it is Jehowah's Witnesses, not Wikipedia!

blog.pangeanic.com
4
5
One of the best summaries of our history @Pangeanic : from our beginnings as a #translation service to our collaboration with @ITI_TIC and @UPV , neural networks, EU projects, #transformation and evolution to #data for #AI
Valencia Plaza
@valenciaplazaSep 14
Pangeanic, de traducir manuales industriales a entrenar la IA de las grandes tecnológicas 🔗 buff.ly/VpMwWJ6 La compañía valenciana trabaja para empresas como Microsoft y Apple en el suministro de paquetes de datos y desarrolla sistemas soberanos para las administraciones públicas
1
98
One striking observation at #LREC 2026 was how rarely the most consequential discussions revolved around building ever-larger language models. Attention repeatedly returned to evaluation, multilingual data, human supervision and deployment. blog.pangeanic.com/what-lrec…

What LREC 2026 Revealed About the Future of Multilingual Enterprise AI

Pangeanic reflects on LREC 2026: multilingual AI, evaluation, trusted datasets, DAAIT, Mozilla Data Collective, ALT-EDIC and enterprise deployment.

blog.pangeanic.com
60
Developing Multilingual AI? Evaluation is the new annotation. Vet partners on quality, coverage, privacy, and safety. See the 7 criteria 👉 blog.pangeanic.com/7-criteri… #AI #DataOps

7 Criteria for Choosing Multilingual AI Training Data Services for Custom Models

Learn how to evaluate multilingual AI training data services through annotation quality, governance, model alignment, behavioural benchmarking and continuous evaluation.

blog.pangeanic.com
46
Japanese forces AI systems to solve problems that rarely appear in English: mixed scripts, omitted subjects, honorific language, complex tokenization and heavy dependence on context. More on our evergreen guide blog.pangeanic.com/japanese-… #Japanese #MultilingualAI #AIData

Japanese Language: History, Writing System, AI Challenges | Pangeanic

Explore the history, writing system and linguistic features of Japanese, and the challenges it creates for machine translation, speech AI, OCR and LLMs.

blog.pangeanic.com
57
Multilingual AI rarely fails everywhere at once. It fails locally. Our new article explains language coverage, tokenization, annotation, evaluation and Local Quality Collapse. blog.pangeanic.com/why-multi… #MultilingualAI #AIData

Why Multilingual AI Data Quality Is Hard to Get Right

Multilingual AI data quality depends on tokenization, annotation consistency, cultural context, provenance, and language-specific balance and evaluation.

blog.pangeanic.com
3
4
56
Pangeanic retweeted
valgrai@fvalgraiJul 14
🎥 Ya puedes ver la Company Talk de Manuel Herranz (@Pangeanic) en el #ValgrAI #VSCF2026, donde compartió la visión de la compañía sobre el papel de las ontologías, los modelos pequeños y la IA soberana en el futuro del desarrollo tecnológico. 🔗 youtu.be/lkIVmZNcPAY #IA #AI
2
4
54
Reliability in an LLM depends on curation, not volume. Moving from Fine-Tuning to Red Teaming requires traceable Data Pipelines, RLHF alignment, and hallucination mitigation under the EU AI Act. The technical infrastructure behind AI Data Operations: blog.pangeanic.com/from-fine…

From Fine-Tuning to Red Teaming: The Data Operations Behind Reliable AI Models

Learn how SFT data, expert reasoning, multilingual AI red teaming, human evaluation and regression testing create more reliable enterprise AI models.

blog.pangeanic.com
1
4
38
After the #Spanish government announcement labelling Palantir “the most dangerous company in the world” it makes sense to read our previous analysis on Captive AI blog.pangeanic.com/tokens-ar…

Tokens are the new coal… for “Captive AI”?

Palantir’s 'tokens as the new coal' means shifting from CapEx to OpEx on cognition (Airbus, BP, US Army), leading to the ineluctable need for Sovereign AI

blog.pangeanic.com
1
4
40
After official presentations, Virginia Dignum from @umeauniversity kicks off ValgrAI @fvalgrai Scientific Council Forum 2026 with some statistics on how #data provenance skews AI models towards certain cultures and coun
1
3
163
Do you wish you could move beyond "fixed machine translation" from APIs and have your own model that follows your style, that adapts terminology... in seconds? Meet Marina Albert Girona and Deep Adaptive at #LREC! #AITranslation #DeepAdaptive #machinetranslation
1
61
Riccardo Brambilla gave Pangeanic 5 stars via @Trustpilot trstp.lt/_6FUOnp6WX
1
21
A few things Pangeanic's ECO Intelligence Platform can do but ChatGPT Translate can't bit.ly/4vnBTyl
3
79
En Pangeanic nunca pedimos pagos o datos por WhatsApp ni apps de mensajería. Si alguien haciéndose pasar por alguien del equipo te escribe, no respondas. Verifica por canales oficiales: [email protected] bit.ly/4djko9H
59
🎥 AI4Culture Final Conference recordings are live! Our CEO, Manuel Herranz, & Mª Ángeles García Escrivá joined experts to discuss how AI is transforming cultural heritage: 🔹 Metadata enrichment 🔹 Machine translation 🔹 Automatic subtitling Watch here: bit.ly/4i5Jw4F

AI4Culture

The AI4Culture Conference hosts researchers and professionals that will showcase real-world applications of AI in cultural heritage. From metadata enrichment to content analysis and automatic...

vimeo.com
1
51
AI: Friend or Foe? 🤖 At The AI Summit New York, Manuel Herranz, CEO of Pangeanic, tackles the fear-mongering around AI. Should we be worried, or is it all just noise?
2
3
66
What a futuristic panel tomorrow at @Slator: Bruno Bitter from @Blackbird, Manuel Herranz from Pangeanic on the State of the Art in AI Translation! bit.ly/4hyIkGp
1
55
🚀 Don’t miss Manuel Herranz’s talk on Tuesday the 11th at AI4Culture! Talk: From Hype to Reality: Future-Proofing AI Investments. How can businesses balance AI innovation & compliance? Join us to explore privacy-first architectures & EU-aligned strategies!
39
Load more