Sabi CEO Rahul Chhabra is pitching a beanie he claims can read your mind. The Palo Alto startup, backed by an eight-figure investment from Vinod Khosla, uses 100,000 biopotential sensors to decode neural activity through the skull and scalp. Their goal: translate silent thoughts into digital text at 30 words per minute.
Founder Jason Calacanis is deeply skeptical. On his show This Week in Startups, he demanded a live demo, warning the non-invasive approach has historically failed and calling the tech “the next Theranos.” The technical hurdle is the skull itself, which muffles electrical signals and has limited non-invasive brain-computer interfaces (BCI) to simple commands.
“By using deep learning models to filter signals through the skull, we can achieve high-fidelity transcription without drilling holes in the head.”
- Rahul Chhabra, This Week in Startups
Calacanis is more confident in a different frontier: AI sidekicks for live content. He launched a $5,000 bounty for a real-time AI “sidebar” that fact-checks and roasts a host mid-stream. The leading submission, “Armchair” by Mark Colbruga, uses Deepgram for transcription and Google’s Gemini for analysis to provide cited corrections and snarky commentary on live transcripts.
His criteria have narrowed to prioritize speed and accuracy over humor, seeking a “cynic” persona that can challenge a host’s framing during a Zoom call or interview. The contest has drawn about a dozen submissions, with final evaluations scheduled for May 15.
Calacanis argues the credibility of ventures like Sabi hinges on radical transparency and proof. Until he sees a live, unscripted demo where the beanie accurately transcribes a user’s unspoken thoughts, the multi-million dollar bet remains a high-stakes gamble on a problem that has defied non-invasive solutions for decades.
