Three former Google X scientists just closed a $5.7 million seed round for TwinMind, an AI-powered 'second brain' that passively captures background speech to build personalized knowledge graphs. The startup, valued at $60 million post-money with backing from Sequoia Capital, promises 16-17 hours of continuous audio processing without draining your battery - a feat that required months of hacks around Apple's restrictions.
The funding news broke this morning as TwinMind co-founder and CEO Daniel George revealed how his team spent nearly seven months 'finding hacks around Apple's walled garden' to make continuous background audio capture work on iPhones. The technical breakthrough required building a low-level service in pure Swift that runs natively, while competitors using React Native hit Apple's background processing restrictions.
George's journey from Google X to TwinMind began in 2023 during his role as VP and Applied AI Lead at JPMorgan, where back-to-back meetings inspired him to build a script that captured audio, transcribed it, and fed context to ChatGPT. When colleagues on Blind showed interest but balked at running something on work laptops, he pivoted to a personal phone app that could quietly listen during meetings.
The $5.7 million seed round, led by Streamlined Ventures with participation from Sequoia Capital, values TwinMind at $60 million post-money. Stephen Wolfram, George's former research lab mentor, wrote the first check - marking his debut startup investment. The funding comes as TwinMind reports over 30,000 users, with 15,000 monthly actives split between professionals (50-60%), students (25%), and personal users including George's father, who uses it to write his autobiography.
TwinMind differentiates from AI meeting note-takers like Otter, Granola, and Fireflies by capturing audio passively throughout the day rather than just during scheduled meetings. The app processes everything on-device, deleting audio on-the-fly while storing only transcribed text locally. Users can opt out of cloud backup, and the system works offline with real-time translation across 100+ languages.
The technical foundation traces back to George's Google X experience across six projects, including iyO - the AI-powered earbuds team that recently made headlines for suing OpenAI and Jony Ive. 'Google X was actually the perfect place to prepare for starting your own company,' George told TechCrunch. 'There are around 30 to 40 startup-like projects happening at any given time. Nobody else gets to work at six early-stage startups over two or three years.'
Beyond the mobile app, TwinMind offers a Chrome extension that uses vision AI to scan open browser tabs, interpreting content from email, Slack, and Notion. The startup used its own extension to shortlist interns from 854 applications this summer, opening all LinkedIn profiles and CVs in browser tabs before asking the AI to rank candidates. 'It did a fantastic job - that's how we hired our final four interns,' George said.
TwinMind also launched its new Ear-3 speech model, supporting 140+ languages with a 5.26% word error rate and 3.8% speaker diarization error rate. The model, a fine-tuned blend of open-source models trained on podcasts, videos, and movies, costs $0.23 per hour and will be available through an API within weeks. Unlike the offline Ear-2, Ear-3 runs in the cloud but automatically switches back to Ear-2 when internet connectivity drops.
With the Ear-3 launch, TwinMind introduced a Pro subscription at $15 monthly, offering 2 million token context windows and 24-hour email support, while keeping the free version with unlimited transcription hours. The 11-person team plans to hire designers for UX improvements and build a business development team for API sales, while investing in user acquisition.
TwinMind represents a bold bet on ambient computing - the idea that AI should seamlessly integrate into daily life rather than require explicit interaction. With ex-Google X credibility, Sequoia backing, and early user traction spanning professionals to autobiography writers, the startup faces the challenge of scaling while maintaining privacy-first principles. As AI assistants evolve from reactive chatbots to proactive context gatherers, TwinMind's success will likely depend on whether users embrace always-listening AI as helpful augmentation or invasive surveillance.
