Ambercite AI uses an entirely unique approach to find new and unexpected prior based on an analysis of our database of 100 million patents, 50 million patent families, and 155 million patent citation.
But does it work? Yes is the answer - as will be demonstrated in this case study below.
This case study is the US Patent Trial and Appeal Board (PTAB) case of EASTMAN KODAK CO., ET AL, vs CTP INNOVATIONS, LLC,, IPR2014-00788, published on 13th January 2017. CTP Innovations held a patent US6738155 for the remote printing of PDF documents, and this patent was being challenged by Eastman Kodak, Agfa, Esko Software and Heidelberg. The final written decision, found here, shows that this patent was overturned as lacking inventive step in the light of Dorfman (WO199808176) and three items of non-patent literature.
US67138155 had just three items of prior art cited against it, and had a priority date of 30 July 1999. So how we find more prior art?
Running searches for prior art in Ambercite can be very easy - in this case, we can set up a query like this:
This will, as expected, return 250 results, which can be individually reviewed in our easy to use review box.
But maybe there is a smarter approach. To be relevant prior art, it is likely that a patent will need to refer to the word PDF somewhere in the title and abstract. Ambercite Ai comes with te ability to run keyword filters on the titles and abstracts of the patents found, so we might make this a limitation:
By limiting the results in this way, only one patent was left to review - being WO1998008176, the knockout Dorfman patent. Yet this patent had not been previously cited against US6738155 - hence it is referred to as an 'unknown' citation in the patent review box below:
This, in two simple steps, Ambercite was able to uncover the previously uncited prior art patent that helped to invalidate US6738155. This truly is an AI approach to patent searching - that works.