Silicon Valley’s AI wunderkind launches Underdog, the most private Instinct/Muse competitor yet; plus, name Ex-Ramp engineers raise $20M for platform Melius

This is TechCrunch. This episode is brought to you by Framer. You know what's absolutely terrifying? A bad website. Or worse, one that doesn't work. Avoid that nightmare and use Framer. Framer is the AI website builder that helps creators, teams, and businesses ship production-ready sites faster than ever while getting every detail right. Prompt, inspect, edit, and publish in one place at a whole new pace. So I was new to this whole website building thing, but Framer helped me get set up and running incredibly fast. For instance, I wanted to add a blog to my website. It was super easy. Starting with a basic blog page template, I added some breakpoints and then jumped into customizing my layout, adding my own special touches with Framer's intuitive tools, and bam! I was ready to start letting my thoughts flow. See why thousands are trusting Framer to help build their websites. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer.com slash tcstartups for 30% off a Framer Pro annual plan. That's framer.com slash tcstartups. Rules and restrictions may apply. When self-taught coder Sigal Wen was 17, he moved to Silicon Valley and lived in an AI hacker house with famed AI researcher Andrej Karpathy. While there, he hacked and coded alongside other people who had become the biggest names in AI, like Perplexity founder Eravind Srinivas and OpenAI researcher Noam Brown. He tested early versions of AI tools that would later become well-known, including a chatbot shared by Anthropic co-founder Ben Mann that would become clawed. An image generator from David Holtz that would become Midjourney and what would become OpenAI's GPT-3 and the image generator Stable Diffusion. Prominent investor and entrepreneur Naval Ravikant hired him for AirChat, Ravikant's now defunct rival to the Clubhouse social network. For fun, he figured out how to get GPT-2 running on his Apple Watch. It was a magical time, Wynn told TechCrunch. Now, a Thiel Fellow, the program from investor Peter Thiel that invites young founders to pursue projects instead of college, Wynn launched an invite-only beta of Underdog on Monday, one of the most private AI assistants Silicon Valley has yet to offer. The model runs wholly on-device, meaning the user's data remains on devices they already own, currently Macs and Windows PCs, with Linux, iPhone, and Android versions coming soon. Underdog is powered by Husky, the inference engine when built to run AI models quickly on a user's own hardware. Wendt says it moves less data between a computer's main chip and its graphics chip than rival on-device engines. Underdog has other security features baked in as well, like encrypting the keys to the email and other accounts that users authorize Underdog to access. Underdog is, however, using much smaller models than today's state-of-the-art ones hosted in data centers. It currently uses a 27 billion parameter reasoning model, fine-tuned from QN 3.827b. Wen argues that this model compares favorably to Claude Opus 4.6 in some benchmarks or what was considered top performance six months ago. He says that means it can handle the everyday tasks