Several industry groups are dedicated to making the Internet of Things more secure, yet hardly a week seems to go by without a large-scale hacking attack involving connected devices. Mythic Inc. argues that a new approach is required to tackle the problem.

The San Francisco-based startup today raised a $9 million funding round led by Draper Fisher Jurvetson to pursue its efforts. Mythic offers a novel artificial intelligence system that aims to address what it perceives as the biggest weak point of connected devices: their Internet links. The firm’s technology can reduce or even eliminate the need for web access by providing to ability to handle tasks locally that normally have to be offloaded to a remote data center.

Mythic accomplishes that using a sophisticated deep learning model developed by co-founders Mike Henry and Dave Fick, who both hold Ph.D.s in computer engineering. The software is designed to run not on a graphics processing unit chip like most artificial intelligence applications but rather a specialized chip that implements what is known as processing in memory, or PIM. It’s similar in some ways to the artificial neurons that IBM Corp. is working to develop.

At a high level, PIM combines a processor and a memory device, flash in the case of Mythic’s design, on a single integrated circuit. This arrangement allows data to travel much faster between the two components than in traditional computers where they’re implemented separately. As a result, the system provides what the startup describes as a uniquely high level of efficiency.

Mythic claims that its technology can help increase the battery life of a device by 50 times while performing computations that are currently not feasible to perform at the edge of the network. According to a spokesperson, the startup’s design makes it possible to fit a desktop-grade GPU’s worth of processing power on a device the size of a shirt button.

Mythic is currently working with early adopters and expects to start mass-producing the hardware in the middle of 2018. The startup initially plans to target relatively low-power devices such as appliances and surveillance cameras, but sees its technology eventually finding use in much more sophisticated systems as well. Today’s funding announcement lists self-driving cars among the potential use cases on Mythic’s radar.

Steve Jurvetson of DFJ is joining the board part of the investment. The firm was joined by Lux Capital, Data Collective and AME Cloud Ventures in the round.

Image: Pixabay