Sabi Raises $50M Seed for a Wearable Brain-AI Interface

Reading neural signals through hair gives Sabi a demanding hardware problem before its software receives an instruction. The California brain-AI startup announced a $50M Seed round on October 9, 2026, led by Khosla Ventures, to develop an interface inside an ordinary baseball cap. Co-founders Rahul Chhabra, CEO, and Atmadeep Banerjee, CTO, are building toward intentional communication with AI without typing or speaking.

Accel, Initialized Capital, Kevin Weil, DST Global, Collaborative Fund and Ascend also participated. The capital will support neural-data collection, completion of a custom chip, early manufacturing and specialist hiring. Sabi plans to demonstrate an early prototype at CES 2027 in Las Vegas, with everyday performance still to be established.

The choice of headwear brings consumer behavior into the engineering brief. A useful interface must fit comfortably, interpret an intended message and give its wearer control over what happens next. Sabi's announcement puts recording permissions, application access and action confirmation alongside the work of decoding neural activity.

What Sabi's Seed Round Funds

Sabi's $50M financing supports a product built across several expensive disciplines. Its announced use of proceeds covers a larger neural dataset, completion of its custom EEG chip, manufacture of early devices and growth of its neuroscience, hardware and machine-learning teams. The company is also preparing units for people on its early-access waitlist.

That allocation helps explain the scale of the development task. A consumer cap has to accommodate sensors, electronics, communications and the ordinary variations of wearing something on the head. Model development depends on the quality of the signals that hardware can acquire. Manufacturing, in turn, has to preserve the properties a prototype team worked to achieve.

Khosla Ventures describes its investment approach as early backing for large problems susceptible to technological solutions, including deep tech and new consumer habits. Leading Sabi's round places capital behind an interface proposition: a person could communicate with an AI system through a wearable that interprets deliberately formed speech or trained commands. The firm's published thesis provides context for that proposition, while Sabi's actual results will have to come from the devices and users.

The announcement does not disclose a valuation, earlier round amounts or a verified lifetime funding total. Keeping the accounting at the announced $50M Seed avoids turning an individual financing into a larger capital history that the source materials do not establish.

From Brain Recordings to Consumer Headwear

Sabi describes a cap built around custom EEG electronics and miniature sensors intended to acquire neural signals through hair, without gel, surgery or direct scalp contact. Its official product site shows cap, beanie and bucket-hat concepts. The October funding announcement centers the baseball cap and a planned early demonstration at CES 2027.

The company says its Brain Foundation Model has been trained on more than 100,000 hours of labeled, non-invasive recordings. Its planned interactions span imagined speech becoming text, trained mental commands becoming AI prompts, and predicted keystrokes or actions presented for completion or confirmation. These are company-described capabilities under development. Dataset volume alone gives readers no measure of how reliably the complete cap will work for a new wearer.

Banerjee brings an independently documented research background to the effort. He co-first-authored MindEye, published at NeurIPS 2023, which studied retrieval and reconstruction of viewed images from fMRI recordings. The paper is useful evidence about the CTO's experience connecting neural measurements with machine learning. Its imaging modality and task differ from the EEG imagined-speech interface Sabi is developing.

Moving into headwear adds its own product requirements. Comfort, fit, power consumption, repeatable signal acquisition and understandable feedback all have to coexist in something people can choose to wear. The funding gives Chhabra and Banerjee resources to work across that whole chain, including the pieces a software demonstration can leave offscreen.

The Wearer Still Owns the Instruction

Sabi's product design raises a practical question for AI application builders: when should an interpreted neural signal become an action? Imagined speech, a trained command and a prediction of what someone might do are different interactions. An application receiving them needs to preserve those distinctions rather than treat every model output as permission to proceed.

The company says it is developing controls for when the cap records, which applications may receive commands and which actions require confirmation. That design work makes authorization visible as part of the interface. A wearer could need to review a suggestion, correct an interpretation or stop a recording session; how clearly the product supports those choices will shape the experience of using it.

Sabi also describes encryption beginning at the sensor and a system designed to perform inference while neural data remains encrypted. Readers should treat this as the company's intended architecture. The announcement does not establish an independent security audit, a certification or externally reproduced protection guarantees. An eventual buyer will need evidence about implementation as well as a description of the design.

For operators building AI applications, the useful implication is specific. Adding a new input channel changes how software receives intent and how a user can inspect or withdraw it. A system that predicts an on-screen action has to give its application somewhere to put the prediction while the person retains authority over execution.

A Prototype With an Everyday Ambition

Sabi's official site currently invites people to join a waitlist. The funding announcement describes early devices and an upcoming CES 2027 prototype rather than providing a price, a firm commercial shipment date or measured customer adoption. Those missing details place the financing at a development milestone in the company's public record.

The consumer ambition reaches beyond solving a decoding task in a controlled setting. People have routines, preferences and very limited patience for an input device that needs constant attention. A cap intended for ordinary life has to become familiar enough that its wearer understands when it is listening, what it has interpreted and where a command will go.

That is the operating relationship this Seed round finances. Chhabra's team is working to connect deliberately formed words and commands with AI applications; The technical work has to make that relationship dependable. Their early-access users will encounter the entire interaction, from putting the cap on to deciding whether the software understood them.

As Sabi prepares its demonstration and early units, the most revealing moments may be the small ones around a proposed instruction: the correction a wearer can make, the confirmation an application asks for and the choice to keep a thought private. Those moments belong in the product experience the company is now funded to build.