Making a Connected AI Kitchen Camera Commercially Viable

The challenge

A global recipe and cooking platform wanted a small camera that sits in a home kitchen and photographs each step of a dish for its AI cooking lessons. The client's earlier prototype worked, but its bill of materials was more than $120 per unit, far too high for a consumer device sold alongside a subscription. Other manufacturers the client approached said the product could not be built at a viable cost.

What we did

We started from first principles: what image quality the coaching service actually needed, how much on-device computing that required, and how the camera would connect and receive updates. Every component in the earlier design had to earn its place against those requirements.

Much of the saving came from moving work from hardware into software. Custom image pipelines produce the photos the service needs from far less expensive camera hardware, and a custom circuit design removed parts the product did not need. The camera still had to cope with heat, steam and grease, set up easily on home Wi-Fi, and stay secure and updatable in the field.

We carried the product through proof of concept, prototype, engineering validation (EVT) and design validation (DVT). This work draws on our consumer electronics development, IoT development and PCB design services.

The result

The bill of materials fell from over $120 to under $30, a reduction of more than 75%. The camera is now in mass production in China.

Work With Us