Abstract: Simular— “like a human but not human”—is dedicated to advancing AI agents that interact with computers as fluidly as humans, 
aiming at pushing machines toward general intelligence. In this talk, I will unveil Simular’s latest agent design paradigm, which combines System 1 (fast, intuitive) and System 2 (slow, deliberate) thinking. This dual-process approach enables our agents to adapt dynamically to tasks, marking a new era in human-computer interaction. I’ll talk about our production effort on computers and browsers and introduce Agent S, our recently launched open-source System 2 agent, currently top-ranked in the osworld academic benchmark. Join us to explore the future of intelligent digital interfaces.

Bio: Ang Li is the CEO and Co-Founder of Simular, a tech company creating AI agents that can use computers like a human. He earned his Ph.D. in computer science in 2017 from the University of Maryland College Park, specializing in representation learning for linguistic and visual information. Ang has accumulated a wealth of expertise in on-device and large-scale machine learning through his past roles at Google DeepMind, CMU Robotics, Apple Camera, Google Maps, and Facebook AI Research. From 2017 to 2021, Ang led multiple applied research programs at Google DeepMind, driving advancements in evolutionary neural networks and lifelong learning, and contributing to Google’s production 
ecosystems. Between 2021 and 2023, Ang served as Principal Scientist at Apollo, Baidu’s Autonomous Driving Unit, where he spearheaded the 
development of state-of-the-art continual learning platforms for self-driving cars, advancing the frontier of autonomous vehicle technology. Ang published over 39 papers in top-tier AI conferences and served as senior program committee members at top AI venues like AAAI and IJCAI. Ang started to code at an early age, leading to achievements in competitive programming such as first place in the 2013 Mid-Atlantic ACM/ICPC and 39th rank in the World finalist of ACM/ICPC.

Q&A: Submit your questions here: https://forms.gle/bzKEQWtgDA6cr52e6