Frontier AI labs
Force-aware training data for vision-language-action and world models.
Giving robots a nervous system.
Preload pairs everyday video with scalable, synchronized multimodal sensing, capturing how hard human hands hold, press, and let go.
Contact us
Backed by
The problem
Video flattens the physical world
The physical world is a dense cloud of force, contact, and motion. Video captures only what a camera can see: a flat projection of it.
Better model architectures can't recover what was never measured. Each sensor we add brings the data closer to the real world.
Our solution
End to end, from sensors to automation
-
Collect
Capture video with the right sensors for each task.
-
Clean
Synchronize, filter and annotate the data for quality.
-
Evaluate
Train models and measure them against ground truth.
-
Deploy
Turn validated models into automated solutions.
Our hardware
Three signals, recorded in sync
-
Vision
A head-mounted camera records each task from the worker's point of view.
-
Tactile sensing
A sensor glove measures pressure across the fingers and palm.
-
Muscle activation
An EMG band on the forearm reads the effort behind every grip.
Who we serve
Teams whose models have to touch the world
Why Preload
Built by people who decode neural data
Our team has studied and worked at
-
Neural data experts
Decoding high-bandwidth neural signals at leading BCI companies.
-
The full stack, in-house
Collection, evaluation, policy training, and real-hardware testing.
-
Built for real deployment
We start from industrial tasks that have to work, not demos.
-
From labs to factories
Industrial access to real tasks, and a close loop with model teams.
Contact
Get in touch
Help us reinvent robotics inference.