# Arctop

> Arctop is brain-computer interface (BCI) software: it decodes focus, cognitive load and attention in real time from the EEG in everyday headphones, so apps and AI can respond to the person wearing them. Founded 2016, San Francisco.

- **What Arctop is:** brain-computer interface software. It decodes cognitive state (cognitive load, visual attention, auditory attention, enjoyment, sleep, memory and identity) in real time from the EEG sensors in everyday headphones and other wearables, so apps and AI can respond to the person wearing them. Arctop does not make hardware.
- **Custom classifiers:** beyond the seven ready-made outputs, Arctop builds custom classifiers for partners: any state a study can define and measure, served through the same API.
- **Founded:** 2016, San Francisco, by Dan Furman, Ph.D. (CEO) and Eitan Kay (CTO).
- **Investors:** In-Q-Tel, Supermoon Capital, FGF (Fifth Growth Fund) and Script Capital.
- **Patents:** 13 granted patents (U.S. 6, Europe 2, Japan 2, Israel 3), listed at https://arctop.com/data/patents.json.
- **U.S. Air Force:** a 1.25M USD AFWERX Direct-to-Phase II SBIR contract for cognition-adaptive training simulations, with MetaTeq and BigBear.ai.
- **Shipping:** the Arctop app runs on Master & Dynamic's MW75 Neuro EEG headphones (12 channels at 500 Hz) and measures cognitive load live.
- **Developers:** a public API with self-serve keys (https://docs.arctop.com): apps on any connected device read the decoded scores in real time.
- **AI:** Reinforcement Learning from Brain Feedback (RLbF, preprint, August 2026, https://doi.org/10.20944/preprints202608.1274.v1) turns decoded EEG into a reward signal that adapts a language model to the person using it; Isaac, an EEG-adaptive conversation agent, ships in the Arctop app.
- **Privacy architecture:** EEG is encrypted on the device and decoded in Arctop's cloud; language models see derived scores, never raw EEG.

This file points language models at Arctop's primary public content: long-form research posts on BCI, EEG, and brain-decoding methods, plus core company pages. The home page is a WebGL film; the facts above cover what it presents, with public details it leaves out, and the deep dives below carry the substantive material.

## Deep dives

- [Teaching AI to Read the Room](https://arctop.com/deep-dives/rlbf-reinforcement-learning-from-brain-feedback): A plain-language companion to our RLbF paper. How real-time brain signals can teach language models how their words actually land, and why those skills might persist after the headset comes off. (2026-08-23) DOI: https://doi.org/10.6084/m9.figshare.33325116
- [Why Consumer EEG Embedded Hardware Is the Next Big Platform: My Predictions as a CTO](https://arctop.com/deep-dives/consumer-eeg-hardware): A look at the consumer EEG hardware market and why it's poised to become a major platform for developers. (2025-03-28)
- [Memory Decoding at Home: Mapping Inter-Subject Neural Synchrony to Episodic Memory Performance](https://arctop.com/deep-dives/memory-decoding-from-eeg-at-home): Research on mapping neural synchrony to memory performance using EEG data collected at home. Our novel FCCA algorithm outperforms traditional methods in decoding episodic memory from consumer-grade brain devices. (2024-07-15)
- [Five Levels of Explanation (Part II): How Brain-Computer Interfaces Work](https://arctop.com/deep-dives/how-bci-works-part-2): An in-depth exploration of brain-computer interface technology in five levels of difficulty. This is levels 4 and 5. (2023-12-29)
- [Computers Will Soon Read Your Mind](https://arctop.com/deep-dives/computers-will-soon-read-your-mind): Technology will help patients suffering from ALS or strokes. (2023-12-14)
- [Five Levels of Explanation (Part I): How Brain-Computer Interfaces Work](https://arctop.com/deep-dives/how-bci-works-part-1): An in-depth exploration of brain-computer interface technology in five levels of difficulty. This is levels 1 through 3. (2023-12-14)
- [Brain-based Authentication: Towards A Scalable, Commercial Grade Solution Using Noninvasive Signals](https://arctop.com/deep-dives/brain-based-authentication): Research on brain-based authentication using noninvasive signals showing how brain activity can serve as a unique biometric identifier for secure, passwordless authentication systems. (2021-04-09)
- [Modeling The Effect of Background Sounds on Human Focus Using Brain Decoding Technology](https://arctop.com/deep-dives/modeling-focus): Research on how different sound environments affect human focus, showing personalized soundscapes significantly improve concentration compared to music playlists and silence. (2021-04-02)

## Site

- [Home](https://arctop.com/): Arctop's brand film, a scroll-driven WebGL piece: from the signal in everyday wearables, through the platform (seven ready-made outputs, custom classifiers, one API), the use cases (headphones at home and flight training, shipping today; health, learning, work, XR and games, media and identity), to the vision: AI that adapts to the person, and the mission, technology that understands people and answers to them. Not text-crawlable; the facts above cover its content.
- [Deep Dives index](https://arctop.com/deep-dives): Hub listing every long-form post.
- [Developer API docs](https://docs.arctop.com): Reference for the Arctop developer API: recording sessions, score outputs, marker events, and a real-time WebSocket. API keys are self-serve from the dashboard linked on that page.

## Optional

- [Contact](https://arctop.com/contact): How to reach Arctop.
- [Privacy](https://arctop.com/privacy): Privacy policy.
- [Terms](https://arctop.com/terms): Terms of use.