Sabi
California startup building a non-invasive EEG baseball cap meant to turn imagined words and trained mental commands into text and instructions for AI assistants. Raised a $50 million seed round led by Khosla Ventures, announced on 9 October 2026; an early prototype is due at CES in January 2027.
Overview
Sabi was founded in May 2024 by chief executive Rahul Chhabra, who worked on AI research at Stanford, and chief technology officer Atmadeep Banerjee, co-first author of the 2023 MindEye paper on decoding viewed images from fMRI. It came out of stealth in April 2026 with plans for a thought-to-text beanie by the end of 2026. The baseball cap now comes first.
Technology
Sensors inside the cap, built around a custom EEG chip still in development, are designed to read brain signals through hair without gel. Decoding runs on what Sabi calls its Brain Foundation Model, which the company says was trained on more than 100,000 hours of labelled recordings. Sabi is developing three interactions: imagined words to text, trained mental commands to AI prompts, and prediction of likely keystrokes or on-screen actions. The single accuracy figure it has given, up to 77 per cent, comes from a wired laboratory prototype, and it has published no peer-reviewed results.
Latest
Khosla Ventures led the $50 million seed round, with Accel, Initialized Capital, Kevin Weil, DST Global, Collaborative Fund and Ascend also investing. Forbes put the valuation at $600 million, citing a person familiar with the deal, and said the cap would not be available until 2027. Sabi has not confirmed the valuation, given a date when customers can buy the cap or named a price.
Why it matters
Sabi aims at everyday consumer use with AI assistants from the start. Merge Labs, by comparison, describes itself as a research lab with a horizon of decades and says its first products will help patients. An employer issuing the cap to California staff could fall within AB 1883, which from 1 January 2027 bars employers from using AI-enabled workplace surveillance tools to collect employees’ neural data, with exceptions that include safety tools.