Kit Signal Specs Who it's for FAQ Reserve a kit

Research kit · BROADMIND

Lab-grade EEG. Everyday wear.

Eight dry-electrode EEG channels at 250 Hz, with 0.16 µVrms of input-referred noise — and an SDK that stays out of the way.

The BROADMIND Research Kit: a white dry-electrode EEG headband with a flexible arm and eight contact electrodes.
Channels
08
Sampling
250 Hz
Noise floor
0.16 µVrms
Resolution
24-bit
Connectivity
USB-C · BLE

Get the highlights.

  • 01

    Dry electrodes

    Contact without gel. No paste, no prep, no technician — put it on and it settles.

  • 02

    Raw data, whole

    Every sample at full rate, unprocessed, streamed to your machine as you record.

  • 03

    Open SDK

    C++, Python, MATLAB, R, Julia, Java, C# and Rust, on an open-source core. The stack you already run.

  • 04

    Swappable electrodes

    Contacts are designed to be replaced one at a time, so a worn electrode never retires the kit.

Drag to explore →

Three things decide whether an EEG is worth wearing.

01 — Contact

The electrode finds its own pressure.

Eight dry active contacts sit in a self-adjusting arm that follows the skull, each pre-amplified five times before the signal travels anywhere. Contact stops depending on how carefully someone put it on.

02 — Montage

Ten–twenty, over the motor cortex.

CP1, CP2, C3, C4, CP5, CP6, FT7 and FT8 — the sensorimotor set that motor-imagery work is built on. Cz is ground and CPz is DRL, so every recording is comparable with published work.

03 — Readout

Signal you can open anywhere.

Band power, artefact labels and raw streams in one window. Export CSV, EDF or JSON — or take the arrays straight into your own pipeline.

What a lab would ask for.

08 EEG channels
250Hz Sampling rate
0.16µVrms Noise floor · input-referred
24bit Resolution

Full Kit figures as published by BROADMIND: 8 dry active electrodes, 0.02–125 Hz bandwidth, USB-C or magnetic 360 connector, Bluetooth LE, one shell that self-adjusts from S to XL. The Starter Kit uses passive electrodes and is specified at 0.8 µVrms noise floor.

Nothing between the brain and the file.

The 10–20 montage exists so recordings can be compared. BROADMIND keeps eight positions across the band and refuses to hide the rest.

Raw, at full rate.

Delta through gamma is computed on device or on your machine — your choice, because the untouched stream is always one call away.

Artefacts labelled, not deleted.

Blinks and jaw clench are marked in the stream. Nothing is silently removed.

blink1 288
clench214

Opens anywhere.

CSV, EDF and JSON out of the box, so a recording made here loads in EEGLAB, MNE or a spreadsheet.

CSV EDF JSON

An SDK that gets out of the way.

Eight language bindings on an open-source core, so a recording made here loads in the tooling you already trust.

Python C++ MATLAB R

Your data is your data.

Recordings live on the device and on your machine. Nothing leaves unless you send it, and nothing you record is used to train a model.

01

Local by default

The device writes over USB-C. There is no account required to record and no cloud in the path.

02

No training on you

Your signal is not a training set. Aggregates are opt-in, anonymised and documented before they exist.

03

Export or delete

Both are one action, always available, and neither asks you to explain yourself.

One measurement core. Three ways to use it.

Research & development

The hardware, opened up.

Raw streams, an open SDK and a montage that matches published work — for prototyping BCIs without buying an amplifier.

Read the developer notes
Teams & labs

Cognitive load, measured.

Aggregated, privacy-first readings of focus and recovery across a cohort — piloted, never sold as a verdict on a person.

Talk about a pilot
Individuals

Your own baseline.

Learn what your attention and recovery actually look like across a week, instead of borrowing someone else's chart.

Join the first cohort

Questions? Answers.

What comes in a kit?
The headset, a USB-C cable, a quick-start and lifetime access to the open-source SDK with example pipelines. The Full Kit adds active electrodes with five-fold pre-amplification and the AI-SDK; the Starter Kit uses passive electrodes at a 0.8 µVrms noise floor. No gel, no consumables, no dongle.
Do I need gel, or a technician?
Neither. The electrodes are dry and the arm is flexible, so most sessions are a short settle followed by a stable baseline. If a contact is poor the app says which channel and why, rather than quietly recording noise.
Can I use my own analysis pipeline?
Yes. The raw stream is available at full rate in open formats, alongside C++, Python, C#, Java, R, MATLAB, Julia and Rust bindings built on an open-source core. Nothing is pre-filtered in the file you export.
How accurate is it, really?
“Lab-grade” is a claim about acquisition — channel count, sample rate, contact quality and noise floor — not about interpretation. We publish our noise-floor measurements, and the raw stream lets you judge the signal yourself.
Where are the electrodes placed?
Eight active contacts at CP1, CP2, C3, C4, CP5, CP6, FT7 and FT8, with Cz as ground and CPz as DRL — a sensorimotor set aimed at motor-imagery work rather than a full-head survey. The shell self-adjusts from S to XL, and contacts are replaceable individually.
Is BROADMIND a medical device?
No. BROADMIND is a research and performance instrument. It does not diagnose, treat, cure or prevent any condition, and nothing it reports is medical advice.
When do kits ship?
The first batch goes to the research cohort, then to the waitlist in order. Reserve a place to get the schedule before it is public.

Your mind deserves better instruments.

Join the first cohort and help shape the platform before it ships wide.