Biosensing and edge inference for Industrial & Transportation Safety

Safety systems see the behavior. Sabi sees what comes before it.

A camera reads a face that has already gone slack. A lane alert fires after the truck has already drifted. Sabi measures attention and focus at the source and hands your safety stack the minutes it has never had.

The Sabi Cap — a ribbed knit beanie with an inner lining of embedded biosensing dots
Backed by
Khosla VenturesAccel+ leading institutional investors
Covered in WIRED This Beanie Is Designed to Read Your Thoughts
The Opportunity

Move your safety signals upstream.

Inference from behavior can only begin once the behavior begins. Sabi starts one step earlier.

Industry Standard
Hours-of-service logs tell you how long someone has been working.
What Sabi adds
A measured read on whether they are actually alert right now.
Industry Standard
Cameras infer drowsiness from eyelids, head pose, and gaze.
What Sabi adds
The neural signal those systems are approximating, measured directly.
Industry Standard
Wearables that come off — wristbands, vests, AR headsets abandoned within the hour.
What Sabi adds
A hat. Head-worn compliance is already enforced on your sites.
Industry Standard
Fatigue reconstructed after the fact, in an incident report.
What Sabi adds
A timestamped, continuous record — usable in prevention and in defense.
The platform

One cap. A full sensing stack.

CapabilityStatusWhat it does
Attention & focusLive todayScores from 0–100, derived from gamma, beta, alpha, theta, and delta activity captured by a high-density EEG electrode array.
Biosensing — neuralLive todayNon-contact electromagnetic sensing, delivered as a sealed package inside the cap. No gel, no prep, no setup by the wearer.
Biosensing — cardiovascularComing soonPPG-based heart rate and blood-oxygen (SpO₂) sensing.
Biosensing — motion & temperatureComing soonInertial measurement (IMU) for movement, impact, and fall events, plus separate body and ambient temperature — the pairing that makes heat exposure legible.
On-device computeComing soonEdge NPU processes sensor data in real time on the wearer's head, and handles command and control of on-device subsystems.
AudioComing soonOpen-air audio above the ears for assistant, comms, and two-way calls, with echo cancellation and noise suppression. Ears stay unobstructed.
Context cameraComing soonConcealed low-frame-rate camera for scene context to the on-device assistant — recognizable scene content, not photography.
ConnectivityComing soonBLE and Wi-Fi: phone link plus direct-to-cloud, so the cap reports with or without a paired device.
PowerComing soonAll-day runtime on two pocketed battery packs with USB-C charge. Full-shift coverage without a mid-shift dock.
Swappable outer shellComing soonUser-swappable cosmetic shell over a soft electronics-bearing inner core — so it becomes the hard hat liner, the cap, or the beanie your workforce already wears.
Where it lands

Built for the operators who carry the risk.

Fleet & Transportation

Hours-of-service tells you how long a driver has been behind the wheel. Sabi tells you whether they are alert enough to be there, turning fatigue from a scheduled assumption into a measured signal you can act on before an incident and point to after one.

  • Carriers and private fleets
  • Telematics and safety platforms
  • Truck and vehicle OEMs
  • DMS validation and L2/L3 handoff research

Construction & Field

Your crews already wear a hard hat on every job site. Sabi turns it into an instrument via continuous attention, cardiovascular, heat, and motion sensing. Fatigue, heat stress, and falls surface in the moment rather than in an incident report.

  • General contractors
  • Heavy equipment operators and ground crews
  • Heat illness prevention programs
  • Man-down and fall detection

Warehouse & Industrial

Labor systems optimize against throughput because throughput is the only thing measurable today. Sabi adds the other half — a real-time read on cognitive load, plus a hands-free assistant running on-device, so work can be sequenced to the capacity a workforce actually has.

  • WMS and labor management vendors
  • Industrial AR and remote expert platforms
  • Hands-free picking and guided work
  • Full-shift wear time, no headset
How it deploys

It joins your stack. It doesn't replace it.

01

Sense

High-density electrode array reads neural activity continuously through the cap. Nothing for the wearer to attach, adjust, or remember.

02

Process at the edge

The on-device NPU turns raw signal into attention and focus scores on the wearer's head. Raw neural data never has to leave it.

03

Transmit

Scores and events move over BLE to a paired device, or direct-to-cloud over Wi-Fi where no phone is in the loop.

04

Integrate

Delivered to your telematics, safety, or labor platform as a scored input — feeding the workflows and thresholds you already run.

Worker trust

A safety program only works if the workforce accepts it.

We designed the architecture around that constraint, not around it as an afterthought. Deployments that read as surveillance fail — in works councils, in union shops, and in daily wear rates.

Edge-first by design

Inference runs on the device. What leaves the cap is a score and an event, not a raw neural stream — a technical property of the system, not a policy promise.

Scored, not surveilled

Sabi outputs attention and focus states relevant to safety. It is not a productivity ranking tool and we don't sell it as one.

Wearable people keep on

It's the hat they already wear, with a swappable outer shell. Wear-time is the metric every wearable safety program dies on, and it's the one we built for.

Find out what your safety stack looks like with a few minutes of warning.

Book a time with us and we'll walk through what a pilot looks like for your team.

Book a pilot conversation
The Sabi Cap — a ribbed knit beanie with an inner lining of embedded biosensing dots