Sabi's brain-reading beanie wants to replace the keyboard

The mechanical keyboard is 150 years old. To finally kill it off, Elon Musk's Neuralink wants to surgically drill a hole in your skull. A quiet Silicon Valley startup thinks you just need to put on a warm hat.

Startup Sabi stepped out of stealth to unveil a non-invasive brain-computer interface disguised as an everyday beanie. CEO Rahul Chhabra says its first brain-reading cap is slated to arrive by the end of the year, with a baseball cap version also in the pipeline.

Billionaire venture capitalist Vinod Khosla, an early backer of OpenAI whose firm invested in Sabi, believes this is the only realistic road for mind-reading tech. If you want a billion people to control computers with their minds, Khosla argues, it simply cannot involve brain surgery.

Packing 100,000 sensors into knitwear

Non-invasive BCI usually relies on electroencephalography (EEG) sensors sitting on the scalp. The fundamental problem is that human skull and skin dampen neural electrical signals down to a faint whisper, limiting most wearable rigs to binary commands like moving a cursor left or right.

Sabi's answer to this signal degradation is sheer density. While typical research EEG setups feature a few dozen to a few hundred electrodes, Sabi packs anywhere from 70,000 to 100,000 miniature sensors into its headwear. The company claims this microscopic sensor grid can pinpoint exact neural activity in real time, capturing the subtle electrical signatures of imagined inner speech.

Reading the signal is only half the battle. Because everyone's neural firing patterns look wildly different even when thinking the same sentence, Sabi is building what it calls a brain foundation model. Trained on 100,000 hours of continuous neural recordings from 100 volunteers, the AI aims to generalize across users out of the box.

Sabi is targeting an initial output speed of roughly 30 words per minute. That is noticeably slower than average physical typing, though Chhabra claims recognition will speed up as the model tailors itself to individual wearers over time.

The out-of-the-box hurdle

Transforming neural hardware into a daily consumer device is notoriously brutal.

JoJo Platt, an independent neurotech consultant based in San Francisco, notes that consumer brain-wearables must operate seamlessly without friction. Most laboratory BCIs require tedious daily recalibration because fatigue, stress, and mood alter brainwave signatures. For everyday consumers, the gear has to adapt to the human instantly, not the other way around.

Form factor matters just as much. Users do not want to look like lab subjects in public, which is why Sabi's beanie and baseball cap disguises are deliberate design choices.

Then comes the privacy nightmare of digitizing human thoughts. Sabi claims all data leaving the beanie is end-to-end encrypted before hitting cloud servers, and its AI foundation models train directly on this encrypted data. The startup is currently consulting neurosecurity researchers from Stanford University to audit its entire software and hardware stack.

Why it matters

We see Sabi's emergence as a defining moment in the ideological split within neurotechnology.

While surgical implants like Neuralink capture breathtaking signal fidelity, their target audience will remain confined to clinical cases for years. Sabi is placing a massive bet on non-invasive brute force. If packing 100,000 sensors can crack silent speech decoding without medical risk, it instantly redraws the competitive boundary for human-computer interfaces.

For regular users, mind-typing at 30 words per minute opens up compelling silent interactions, from discreetly sending messages in quiet meetings to hands-free computing on crowded commutes. It could also become a massive accessibility bridge for individuals with motor or vocal impairments who cannot undergo surgery.

That said, we remain skeptical of the practical hurdles. Maintaining reliable scalp contact through thick hair, managing heat, handling sweat, and figuring out how to wash a garment housing tens of thousands of micro-sensors are non-trivial hardware puzzles. Sabi's vision is clever, but proving it can work reliably on everyday heads remains the ultimate test.