When self-taught coder Sigil Wen was just 17 years old, he moved to Silicon Valley and lived in an AI hacker house with renowned AI researcher Andrej Karpathy. During this time, he collaborated with future leaders in AI, including Aravind Srinivas, founder of Perplexity, and Noam Brown, a researcher at OpenAI. Wen tested early versions of AI tools that would later gain fame, such as a chatbot developed by Ben Mann, co-founder of Anthropic, which evolved into Claude, and an image generator from David Holz that became Midjourney. He also worked on what would become OpenAI’s GPT-3 and the image generator Stable Diffusion.
Prominent investor and entrepreneur Naval Ravikant hired Wen for Airchat, a now-defunct competitor to the Clubhouse social network. For fun, he even managed to run GPT-2 on his Apple Watch. "It was a magical time," Wen told TechCrunch.
Now a Thiel Fellow—a program by investor Peter Thiel that encourages young founders to pursue projects instead of attending college—Wen launched an invite-only beta of Underdog on Monday, which he claims is one of the most private AI assistants available in Silicon Valley.
Key Features of Underdog
The Underdog model operates entirely on-device, ensuring that user data remains on their personal devices, currently compatible with Macs and Windows PCs, with Linux, iPhone, and Android versions expected soon. Wen developed Husky, an inference engine designed for speed, which minimizes data transfer between the computer's main chip and graphics chip.
