Seeing the Future with Clarifai
All the rage now, deep learning and neural networks barely registered as viable startup opportunities in early 2014. VCs weren’t talking about this area back then — the mainstream use cases and applications weren’t yet clear — but we knew the potential of the technology certainly existed, something that stemmed from its long roots in academia.
And that’s when we met Matthew Zeiler, CEO and founder of Clarifai. He applied to Qualcomm Ventures, the investment arm of Qualcomm Incorporated, early stage competition, QPrize, almost three years ago.
He brought with him an impressive reputation and academic pedigree. In the prior year, he had won Imagenet, an annual competition for image recognition technology, viewed as the top honor in the field. He was quietly building one of the most talented machine learning teams in the world out of his New York office. And he was a competitor — we learned that he once scored 77 points in a high school basketball game.
Keep in mind that tech behemoths play in this category. It’s not for the faint of heart, especially for a lean start-up. But Clarifai brought a singular focus on image recognition technology, using an advanced technique called deep learning to identify almost any subject in pictures and videos.
At Qualcomm Ventures, we are comfortable taking early bets on frontier technologies — ambitious and bold ideas tackling non-trivial problems in next-generation platforms and native experiences evolving from the smartphone. In Clarifai’s case, they are demystifying deep learning techniques and unlocking the power of computer vision and AI for the masses. Whether you are a wedding photographer or e-commerce titan, Clarifai can do the AI heavy lifting on your behalf. Clarifai’s API allows any developer to access its technology with seamless “drag and drop” functionality, making the world searchable by photos and videos.
That in itself is a tremendous advancement, but Clarifai’s approach accelerates how we think about what’s possible with APIs. The Clarifai team has productized this in unique ways.
First, Clarifai’s Custom Training feature democratizes computer vision by empowering any developer to train their own models versus simply tapping into a generic one. This represents a significant, positive catalyst as it unleashes much more granular use-cases of AI. In my favorite creative example, a developer applied public images of foul balls in Major League Baseball stadiums to predict the best seats to catch a foul ball. Imagine what this could have done for Steve Bartman!
Second, Clarifai’s Visual Search is an AI tool which arms any subject matter expert or non-tech company with AI capabilities. A technique which started with identifying simple images and video can now empower 3rd parties to moderate offensive content or protect marketplaces from counterfeit goods.
Looking back three years to when we first encountered Matthew and team, we were impressed then and we remain impressed now. We were pleased to participate in Clarifai’s Series A funding in 2015 and again last month in its Series B funding.
Clarifai is pioneering a new and different vision for AI, helping computers see the world like humans — starting with visual recognition. The company’s API has helped analyze millions of pictures for companies around the globe like Unilever, Buzzfeed and 500px.
Broadly, it’s clear that AI is coming of age with startup funding in this sector accelerating (funding for AI startups hitting an all-time high in Q2, according to CB Insights). Large tech players like Google and Microsoft are placing a bigger emphasis on AI and deep learning within their hardware and software. The market is recognizing the significance that AI will have in shaping a new era of computing.
Clarifai recognized this significance back in 2013 and Qualcomm Ventures is excited to continue with the company on its journey toward AI augmenting every part of the human experience.