We believe image similarity is the next frontier in patent search A.I.
I’ve spent the past 16 years in the patent industry. My career in the patent world started at the USPTO, where I was a Mechanical Examiner. Much of my time was spent searching for specific images in the prior art, often leafing through thousands of patents, just looking at pictures. As a private sector patent analyst, I noticed a considerable uptick in design FTO. More and more of our work revolved around comparing images for similarity. This manual process was grueling, and limited. In manual design searching, I noticed the following:
The accuracy of a design search is 1:1 with the number of designs reviewed. Humans can only review limited design images per search, then budget limits or human fatigue kicks in. The more images your search reviews, the more accurate the search.
Companies want to search all of their designs, but can’t afford it
Outsourcing design searching to search firms takes too much time (sad, but true)
Design searching is unique because image similarity is binary in nature. There is no subjectivity. The image is either similar, or not.
The above clearly illustrates a repeatable process that is ripe for automation.
We created Visualize IP (VIP) to solve these problems. VIP reduces cost by 90%, improves accuracy by 500x, and provides instantaneous results. We also achieved something no one has truly figured out in the A.I. patent space - we tell you the “story” of your search. The best human patent searchers are also the best at telling the “story” of the search, providing personalized insights and distilling the landscape into useful pieces. This is how a searcher builds trust in the search result. VIP follows this time-tested human approach. Design patents are just the start. We have plans to expand outside of the patent world. We hope you come along with us for the ride. This is just the beginning.
-Eugene Lhymn, CEO and Founder
