Sabi, which is building a baseball cap that uses 100K non-contact neuro-sensors to convert neural signals into text prompts for AI systems, raised a $50M seed

The cost of using AI assistants that require frequent interaction or complex commands will fall.

Sabi, a startup focused on brain-computer interfaces, has successfully raised $50 million in seed funding. The company is developing a non-invasive "Sabi Cap" that utilizes approximately 100,000 non-contact neuro-sensors and a custom EEG chip to read neural signals. These signals are then processed by Sabi's proprietary Brain Foundation Model to translate internal speech, or the words one thinks without speaking, into text prompts for AI systems. The technology employs electroencephalography (EEG) without requiring implants or scalp-contact gels, aiming to offer an everyday input method. Sabi's approach involves in-house hardware design and fabrication by TSMC, alongside training its foundation model on extensive neural data to reduce individual calibration time. The current iteration targets a word-per-minute rate of around 30, with capabilities for predicting keystrokes.

Sabi's non-contact EEG technology and extensive sensor array aim to overcome the limitations of traditional EEG, which often requires gel and direct scalp contact, making it unsuitable for daily use. By developing its own ASICs and a "non-contact EEG chip" fabricated by TSMC, Sabi is positioning its cap as a seamless, wearable input device for computers and AI tools. This approach could democratize brain-computer interfaces, moving them from specialized medical applications to everyday consumer technology.

The development of Sabi's Brain Foundation Model, trained on a substantial dataset to minimize user calibration, signals a significant advancement in making BCIs more accessible and user-friendly. The company's progression toward thought-to-prompt and intent-to-action capabilities indicates a future where direct neural input could redefine human-AI interaction, potentially impacting how users control AI agents, create content, and interact with digital environments.