Clarifai has traditionally been known as a web-based visual search tool that developers can integrate into their services. But as more and more businesses start to get onto the service, with their own specific demands — like getting them in-house — the New York-based startup has to grow up a bit.
Today, the company is starting a few early steps to do just that. After raising $30 million in October 2016, the company said today that it’s going to start rolling out an on-premise solution for larger enterprises where they can use all of Clarifai’s services in-house without that data shipping outside of their own servers. Larger businesses often have their own kinds of proprietary data and customers to work with, which have myriad different demands, and sometimes have to make sure their customer data doesn’t leave their close watch — or it needs to happen so fast that it can’t afford the lag of connecting to the cloud.
“Think of a casino with 10,000 cameras that all go to a central place,” CEO Matt Zeiler said. “You have like a streaming service that’s [faster than] running through the cloud that needs to run close to the edge. The camera is there, it’s not good enough, so you have to stream it to a central hub. You have to close the latency down, scale it, and protect privacy. It’s all the same API objects and structures you would use.”
The company also said it made two hires at high-level executive spots: a VP of engineering and a VP of product. Ulas Bardak joined Clarifai to head engineering after working at Whisper, while Rajesh Talpade previously served as a product lead for Google. Both of these hires are part of that growing-up process as it starts to try to woo more businesses — and the kinds of talent it needs to build these tools.
Part of Clarifai’s pitch is that it can be a neutral party when it comes to image recognition. While companies like Google and Pinterest can have their own sets of data and build custom visual search algorithms for their products, Clarifai works to build those tools that other companies can tap in order to have their own kinds of visual search tools. It’s part of a process to hand off some of these harder processes to a third party and focus on the key elements of their products, a hallmark of web development that led to an explosion in the startup ecosystem.
“When you use something like us, our research team is continually improving algorithms, and continuing to collect data, and that’ll give you a better level of accuracy and improve over time,” Zeiler said. “That’s a huge advantage and it gets away from how fast you train. We want it to be continuously improving. That takes a lot of resources and data, but that’s the big advantage for signing up for a service like us. We’re always comparing them in terms of quantitative elements like accuracy and qualitative elements like the team that reviews the models before they get published to make sure they’re not making stupid mistakes.”
To cap all this off, Clarifai also is making a bit of a rebrand of the website. That’s a small touch, but it all adds up to ways to help articulate to businesses exactly what they do and that they are building the tools necessary to become a go-to tool for developers when it comes to visual search.
Clarifai, of course, stands at constant risk from some of these larger players and the chance that they’ll just open up their visual search algorithms for developers. There are obvious candidates, but at the same time, if they are trying to build a walled garden they probably want to keep that data and those tools to themselves. If that remains the case, it may buy Clarifai enough time to collect enough data, have all the elements of any given image or video marked up and offer a compelling enough case for businesses to adopt it.