The trajectory of Sigil Wen reads like a modern Silicon Valley odyssey, beginning with a seventeen year old arriving in San Francisco with little more than ambition and a willingness to sleep at a WeWork. By skipping college to embed himself in the city’s most intense hacker houses, Wen positioned himself at the epicenter of the current AI revolution. He spent his formative years collaborating with the architects of DALL-E and GPT 3, eventually becoming an early tester for Claude and receiving mentorship from figures like Andrej Karpathy and Naval Ravikant. This immersion gave him a front row seat to the frontier of artificial intelligence long before it became a household conversation.
Now, Wen is channeling that experience into Conway Research and its flagship product, Underdog. The project is born from a deeply personal motivation involving a past security breach where Wen’s private emails were exposed due to an app bug. That experience instilled in him a conviction that the most powerful AI tools should not require users to surrender their privacy to a cloud server. Instead, Underdog aims to bring high level personal intelligence directly onto user devices, allowing the software to handle sensitive documents and schedules locally and securely.
The technical ambition behind Underdog is significant, moving away from centralized servers toward local execution on phones and computers. Having previously pushed boundaries by running GPT 2 on an Apple Watch, Wen believes that as hardware improves and models shrink, true digital assistants can operate entirely offline. This approach ensures that the data making an assistant useful remains under the sole control of the owner, turning AI into a tool for individual sovereignty rather than corporate surveillance.
Investors leading Conway Research’s first funding round cite Wen’s unique ability to build lasting relationships as much as his technical prowess. From recruiting elite engineering teams for Airchat to being named the youngest member of Spearhead, Wen has developed a reputation for resourcefulness and reliability. By combining these networking skills with a vision for decentralized AI, he is attempting to scale Underdog into a liberating equalizer that provides billions of people with capable assistance without compromising their private lives.
