Andrew Dai

Andrew Dai is the co-founder and CEO of Elorian AI, a multimodal reasoning research and product lab based in Palo Alto, California. Prior to founding Elorian AI, Dai spent more than a decade at Google Brain and DeepMind, where he co-led data efforts for Gemini, a team that expanded to over 400 contributors. During his tenure, Dai collaborated with leading figures in artificial intelligence, including Jeff Dean, Geoff Hinton, Quoc V. Le, Ilya Sutskever, and Dario Amodei. Dai attended the University of Cambridge, and his academic and research background has contributed to his status as a highly cited scientist in language models, machine learning, sequence models, and artificial intelligence, with more than 76,000 citations.

Dai's research has influenced the development of modern AI systems. He is recognized for his work on pre-training and fine-tuning techniques that contributed to the advancement of large language models. Dai co-authored early papers on sequence learning and mixture-of-experts architectures, which have informed the evolution of models such as GPT, GLaM, and ChatGPT.

Elorian AI was co-founded by Dai and Yinfei Yang, formerly chief research scientist on Apple's machine-learning team. The company focuses on developing native multimodal models capable of reasoning directly across images, video, audio, and text, rather than converting visual inputs into text labels. The central thesis driving Elorian AI is the "visual AGI gap," which Dai describes as a fundamental limitation in current frontier models' ability to count, match shapes, and reason over visual information. Dai has publicly stated that progress in visual understanding and visual reasoning remains uneven.

Dai secured a $55 million seed round for Elorian AI at a $300 million valuation, with backing from NVIDIA, Menlo Ventures, Altimeter, Striker Venture Partners, and Jeff Dean. He prioritized strategic investors with deep expertise in artificial intelligence over higher valuation offers. Dai has become a recognized voice in the AI community, appearing on Bloomberg to discuss Elorian AI's progress and participating in the TechCrunch "Build Mode" podcast and the "Inside the Silicon Mind" series. In these appearances, he shares perspectives on communicating complex technical concepts, the importance of speed in AI development, and strategies for recruiting researchers from major technology companies.