Good Enough Isn't Good Enough: Scott Kelly on AI and Patent Quality

In this episode of IP Innovators, host Steve Brachmann sits down with Scott Kelly, Founding Partner and Head of Practice at KellDann, to explore what happens when a patent attorney decides that "good enough" AI-drafted patents aren't good enough. Drawing on nearly 17 years in the field, including three years as a USPTO examiner and 13 years at Banner Witcoff, Scott brings a practitioner's skepticism to the AI boom reshaping his industry.

Scott walks through the founding story of KellDann, a firm he built from scratch with longtime mentor Ross Dannenberg specifically to be AI-forward from day one, and why that's proven more nimble than retrofitting AI onto an existing practice.

But the real thread of the conversation is what he calls "AI slop": the flood of thin, LLM-generated patent applications he expects to flood the system, and why he thinks a lot of them will be vulnerable to abstract idea and written description attacks down the road.

His AI approach is to spend the time up front, often over an hour on a single disclosure call, building the roadmap himself before handing anything to the AI to execute.

‍

‍About the Guest

Scott Kelly is founding partner and head of practice at KellDann, an AI-forward patent firm he co-founded alongside Ross Dannenberg. A former USPTO patent examiner and 13-year veteran of Banner Witcoff, Scott has been recognized as a top PTAB attorney by Managing Intellectual Property, with a specialty practice in video games and interactive entertainment law.

‍