Mundo AI

Mundo AI, established in 2024, is a San Francisco-based seed-stage company addressing the scarcity of high-quality, non-English training data for artificial intelligence models. The firm was founded by Jason Liao (CEO), Kenneth Wu, Naijide Anwaer, Garreth Lee, and Ahnaf Muqset Haque. The founding team possesses a diverse background in quantitative finance, AI research, machine learning engineering, and product management at prominent organizations like Amazon Web Services, Binance.US, Cohere, and Hugging Face. The idea for Mundo AI originated from CEO Jason Liao's personal experience during his AI research abroad, where he encountered the significant difficulty of sourcing non-English datasets, which limited his peers to developing models primarily for the English language.

The company's core business is building a comprehensive multilingual data library to assist AI labs and businesses in developing more proficient non-English AI models. This service is critical for a market where 75% of the world does not speak English, yet AI usability is predominantly English-focused. Mundo AI operates by setting up end-to-end data operations in various countries, working directly with native speakers to create novel datasets. Revenue is generated by providing foundational model labs with off-the-shelf datasets and on-demand data labeling services.

Mundo AI's platform offers proprietary audio and text data, including over 14,000 hours of multi-channel conversations across more than 15 languages. The audio data is recorded on studio equipment by professional voice artists and includes rich metadata on speakers, accents, dialects, and conversation topics. A key feature is the human-led quality control, where every hour of audio is checked, and transcripts are created by native speakers. The company also provides custom audio datasets with specific controls for background noise, tone, and emotions, as well as multi-modal data labeling for video, audio, and text. As a participant in the Y Combinator Winter 2025 batch, Mundo AI secured $500,000 in a seed funding round in February 2025.

Keywords: multilingual data, AI training data, natural language processing, data labeling, machine learning datasets, non-English AI, audio data, speech models, Y Combinator, AI development, data annotation, linguistic data, native speaker data, custom datasets, multi-modal data, AI labs, data-as-a-service, cross-lingual models, language technology, AI infrastructure, computational linguistics, text data, voice data, data collection