By Dan Sanchez | Sep 24, 2018
San Francisco, California
Computing hardware and software
Industry: Electronics & Aerospace
Products: Computing hardware and software
Cousins and his co-founders have created a way to deliver high-end computing power in a small package that's both affordable and accessible.
"The idea started while we were working at a previous firm," says Cousins. "As mechanical and electrical R&D engineers and software developers, we realized many smaller companies had huge amounts of data that required intensive compute power. Unfortunately, this meant using very expensive computers and servers that were typically only available for high-end industrial applications. Working around the problem, we found that many of the medical and industrial clients we consulted for ended up requiring the same core parts, so we decided to turn that into a product."
That product is the snickerdoodle, a circuit board with a special software-reconfigurable processor, hardware peripherals, and software that can perform computationally intensive functions within a compact, low-power, affordable package. "The snickerdoodle board accelerates complex algorithms and is capable of addressing complex applications like artificial intelligence, in a small piece of hardware," says Cousins, likening the snickerdoodle to the small and affordable Raspberry Pi computer for more advanced applications. "It's user-programmable and we can customize the hardware via software, allowing us to change how the hardware is performing these functions."
The business card-sized snickerdoodle can wirelessly integrate with a company's existing systems. "One example of its use is with a company designing a product intended for use in broadcast video," says Cousins. "They would take HD video feeds from in-studio cameras and encode and compress the video in 'real time,' then send it out over the network in order to combine video streams. Our solution was to take the sophisticated compression software that usually operates on larger desktop machines and servers, and, using snickerdoodle, turn it into a device that's the size of couple boxes of playing cards with an Ethernet connection."
According to Cousins, more industries are looking to utilize artificial intelligence to work with the terabytes of data produced by automation, networking, and computer vision, as it can accelerate computation of algorithms as much as 50 times the speed of conventional microprocessors. "We provide hardware, software, and engineering consulting services that help businesses and engineers build great products and get the compute power they need to grow and stay ahead of the curve."
Manufacturing is a big target for krtkl. The snickerdoodle can allow plant operators to remotely monitor robotics and processes in real-time as a key cog in the emerging Industrial Internet of Things, or IIoT. The payoff comes in the form of reduced downtime and increased safety and efficiency.
All of the company's research and development takes place at their San Francisco facility, while much of the manufacturing of the hardware is done by local contractors. "The assembly process is all automated," says Cousins. "And while all the assembly is done in California, the raw circuit boards are made offshore. Due to regulations, existing infrastructure, and other operating costs, it's typically much more cost effective to have circuit boards overseas."
Challenges: Catering to a diverse market; some customers buy a single snickerdoodle, while others buy thousands. "We had a crowdfunding campaign that was successful and we're trying to fulfill our responsibilities with that first," says Cousins.
Opportunities: New products and markets. "The next generations of the product will be complementary to many of the other industries we're getting into," says Cousins. "Our next step is starting to build solutions that are more application-specific. We're creating peripherals that are helping us get closer to the applications that we're now only exposed to at the lowest level. We're also excited that the latest generation of chip technology we're using is well-suited for higher-end applications. We're taking advantage of new and existing partnerships and getting closer to what people will do with it."
Needs: "The capital to get more people," says Cousins. "Beyond that, we need more customers and we're looking into getting into more distribution channels.