AI training dataset supplier specializing in high-quality language data for Nordic and Central European languages.
Letrum Linguistics is a data-for-AI startup focused on a narrow but strategically underserved corner of the AI data market: producing evaluation datasets, preference data, and training corpora for Nordic and Central European languages that are often overlooked by larger AI labs and data providers. The company supplies high-quality annotated language data designed to improve LLM and ASR performance in language pairs where general-purpose multilingual models consistently underperform. By concentrating on specific regional language markets, Letrum positions itself as a specialist supplier for AI teams building or fine-tuning models for Scandinavian and Central European deployments. Its datasets cover text, speech, and evaluation benchmarks aligned with industry standards such as MQM and human preference annotation frameworks used by frontier model developers.