Similari is betting that harried inventors still don't want to play the patent equivalent of "Where's Waldo." The Tel-Aviv based company brought its patent analysis service to the U.S. this week following its first-place finish in the Israel Patent Office's test of automated search tools.

In the U.S., Similari is hoping to set itself apart from other entries in the market by focusing on the ideas in a patent document instead of the actual words.

What is is: Similari is an automated patent search tool that leverages AI and natural language processing to search through the hundreds of thousands of published patents to determine whether or not a new invention falls too close for comfort.

The idea is to give companies a mechanism by which to avoid sinking vast amounts of capital or resources into a concept that has already been spoken for down at the patent office.

Udi Cohen, Similari's CEO, indicated that the company is not looking to totally break the wheel.

"We're not generating a new market, we're just helping the existing market to become more efficient and more proficient with their search," Cohen said.

Why it's Necessary: Per Cohen, the number of published patents that inventors and patent offices have to sort through continues to grow steadily. Plus, two separate entities are apt to use any number of different phrases or words to describe virtually the same idea, which can make accurately executing a fruitful patent search challenging.

Therefore, Similari's target demographic occupies both ends of the spectrum—companies who want to safeguard their investment, and patent offices charged with making sure that new ideas don't bear uncomfortable resemblances to ones that have already been registered.

Under the Hood: Similari's service relies in no small part on artificial intelligence (AI), which means that Cohen and his co-founders Ori Calvo and Tommy Sandbank had to train the tool how to parse a document for content and ideas instead of just words.

But how? The team used a database of past search results that a human examiner had found to be similar to one another so that the AI powering their tool could learn to make the same distinctions.

Just because the tool is active now doesn't mean it will stop learning. Cohen envisions a future beyond patent documents where the tool could potentially be used to compare literature.

"On the database side, we will of course continue to integrate more and more data to make it more comprehensive and able to find documents in places that are even harder to search for," Cohen said.

The Competition: Where there's a problem, there's usually a crowd of people willing to solve it in exchange for cash not far behind.

Similari's competitors include the visual search engine platform DesignVision, which allows users to drop images into a search field and cast a net for similar works. Meanwhile, the InnVenn platform offered by a startup called Loci uses Venn diagrams to help visually inclined overlaps in patent searches.

Also, while not strictly a patent search tool, the law firm of Kilpatrick Townsend & Stockton did make its KT Extranet portal available to clients earlier this year, allowing them to monitor docket information surrounding issued or pending patents.

Similari, meanwhile, is banking on securing a reputation for accuracy to help it remain vital and relevant in the marketplace.

"From the technology side, we're constantly benchmarking ourselves and improving our accuracy…  It's something that the market needs. It's a real problem that we're solving," Cohen said

|