‘CaaS’ Moves Cognition Toward the Mainstream

If you’re among the crush of people at CES in Las Vegas, keep an eye out in case Yi Li has a camera pointed at you. Her startup Orbeus, an API platform company for recognition software, is launching new features for its ReKognition API product that produces not only facial recognition, but scenes. Eventually, it will recognize landmarks and logos.

Orbeus Facial RecognitionLi’s Mountain View, Calif., company offers recognition APIs and SDKs to third party developers. It’s one of the increasing number of businesses starting to emerge in the area of Cognition as a Service, or CaaS.

Companies in everything from consumer products to healthcare to retail want to know more about their customers, business partners, competitors and employees. It’s reasonable to assume they may want to incorporate cognition APIs into their technology, rather than build them in-house.

What does cognition software do? Here’s an example: It may recognize that a consumer smiles every time they see a teal-green car, enabling a system to deliver personalized ads featuring like-colored cars or teal-green upholstery.

“The definition of CaaS is very new,” says Eddy Lee, a principal at Fenox Venture Capital in San Jose, adding that it’s likely to be used by companies and startups that don’t want to have to “reinvent the wheel.”

This CaaS, Not That CaaS

The acronym CaaS has been around since at least the late 2000s, but used in the context of Communication as a Service, which refers to a company that is responsible for handling all the hardware and software needs for multiple customers’ communication services.

But in November, technology futurist Nova Spivack applied the term to Cognition as a Service and declared that CaaS “is the next operating system battlefield.”

The intelligence that powers cognitive apps will come from cloud-based platforms that host their brains — the apps themselves won’t really have to be that smart on their own. Which means that truly vast, always increasing, intelligence will be available via APIs to all kinds of apps, and right into the full range of consumer appliances, devices and even the Internet of things. All apps and even things will start to become cognitive.

CaaS’s Potential

While Spivack largely addresses the use of CaaS for spoken language, Fenox’s Lee says its use can be much broader and apply to information the body itself produces. “For example, if you are running and wearing a wristband with sensors, it could talk to your phone. That is too much information for your phone to process, so it is uploaded to the cloud where all that information is processed. It would then send back information to your phone to change your behavior, like you should drink water now,” Lee explains, adding, “I have only seen a few companies that process cognitive data.”

Fenox has put money into this emerging field, investing in companies such as Jetlore, a Sunnyvale, Calif., company that analyzes social media used by its clients’ customers to understand their personalities and buying preferences. For example, based on information a user enters into Facebook or Twitter, retailers could email targeted shopping recommendations.

Like Jetlore, Orbeus uses machine-learning algorithms and processes information in the cloud to have their APIs issue a call to action, recommendation or other prompt to the end user. Orbeus, which announced its API platform in August 2012, currently has 3,000 third-party developers using its product.

Although Orbeus is getting interest in its public API platform, approximately 90 percent of its revenues come from companies that prefer having it build custom CaaS solutions, Li says. These customers make the choice based on their desire to take extreme precautions in keeping their customers’ information private.

Post a Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>