AI, Invention Harvesting and the Patent Backlog

This week on IPWatchdog Unleashed I speak with my friend Jason Harrier, former Chief Patent Counsel at Capital One and current co-founder and General Counsel of artificial intelligence (AI) company IP Copilot. 

One of the things IP Copilot enables is the use of AI to streamline invention harvesting, running in the background as engineers and scientists engage in their daily activities communicating with each other. So, with this in mind I started our conversation by asking Harrier about invention harvesting, which I know from many conversations with in-house attorneys is one of the more difficult but critical important aspects of their job. We begin with a simple question: Why is invention harvesting so difficult and why are in-house attorneys always talking about in terms of what they tried in the past, what they are currently trying and what they hope to try in the future, sounding a little like Goldilocks looking for what is just right, but always out of reach.

And from there we dive into a conversation about how long scientists and engineers will typically stay with the company—his answer was 18 to 24 months—which led us to broader conversation about how many in the younger generation see work not as a job with normal hours, but a project that once done allows them to move on to whatever may be next. I called this a subcontractor mentality and wondered what this means for the future. 

We then pivot to the patent backlog—which is approximately 860,000 unexamined applications—and whether AI could help the U.S. Patent and Trademark Office (USPTO). According to Harrier, AI tools available today could help patent practitioners and patent examiners get on the same page and streamline patent prosecution. 

This led us into a conversation about patent prosecution strategy, overly broad claims, and how AI could contribute to streamlining patent examination through better prior art searches and with AI identifying the most relevant prior art is, which can often be quite different than the prior art applied by an examiner. If patent practitioner and patent examiner could work together with a closed universe of the most relevant prior art found by AI the ultimate output should be better, strong, more reliable, more difficult to challenge, and the entire process should be faster and more efficient. 

Visit us online at IPWatchdog.com.

You can also visit our channels at YouTube, LinkedIn, X, Instagram and Facebook.