Where it runs
Solutions
From a server down to a chip with no operating system.
01
Servers
The machine a company already runs. A model on it does the repeated work that is sent to an API today.
- Rack servers and workstations
- Virtual machines in a private cloud
- The CPU, no accelerator
- Air-gapped machines
02
Consumer devices
The machines people already own, with enough memory to hold a real model and no good reason to send anything away.
- Phones and tablets
- Laptops and desktops
- Televisions and set-top boxes
- Car head units
03
Cameras, drones, robots
Video is far too big to send and far too useful to throw away. Read it where it is captured and send the sentence, not the footage.
- Fixed inspection and security cameras
- Drones and survey aircraft
- Mobile robots and forklifts
- Agricultural machinery
04
Wearables and sensors
Small enough to sit beside the sensor, and cheap enough in power that the radio can stay off.
- Watches and fitness bands
- Earbuds and hearing aids
- Medical patches and monitors
- Toys and consumer sensors
05
Microcontrollers
Kilobytes of memory, no operating system, and one battery for the life of the product. This is where nothing else runs at all.
- Utility and industrial meters
- Vibration and condition sensors
- Logistics and cold chain tags
- Appliance and equipment controllers
01
No GPU
A CPU, and at the bottom of the range a microcontroller, is enough. The hardware is already inside the product.
02
No inference bill
Running the model costs whatever the device already costs to run. There is no price per request that grows with how much the product is used.
03
Nothing is uploaded
The data is read where it is produced. Nothing has to be trusted to a third party, because nothing is sent to one.
04
No connection needed
The device keeps working in a tunnel, in a field, on a plane, and on a factory floor with no signal.
Work with us
Looking to run our models on your own hardware? Get in touch and we will work out what we can do together.