Kelsey Diffley, Licensing Specialist

The University of Georgia’s Technology Transfer Office (TTO), Innovation Gateway, evaluates invention disclosures from researchers across the university and helps determine whether patent protection is viable. This often requires fast, high-quality prior art searching across a broad range of technical domains from biosciences to engineering and physics.

With a PhD in Chemistry, Kelsey Diffley joined Innovation Gateway through a Postdoctoral training program. Like many postdocs entering tech transfer, she brought deep scientific expertise, but had limited experience with patent search workflows and Boolean querying, and needed to quickly learn to understand and identify relevant prior art documents for a range of different technologies. 
‍

Before adopting NLPatent, patentability searching at Innovation Gateway required working across multiple tools, running similar searches in several platforms, navigating keyword limitations, and relying heavily on Boolean logic to ensure each search was as exhaustive as possible.

For a postdoc without formal IP training, this created two major bottlenecks:

1. A steep technical learning curve
   Searching patent data effectively hinged on knowing the right terminology and Boolean operators, skills that typically take significant training and repetition to master.
   ‍
2. A time-intensive workflow with uncertainty in the results
   When invention disclosures fell outside Kelsey’s core domain, such as inventions rooted in physics or advanced engineering, building reliable search logic became even harder. Historically, a single search could take a full day or more, and there was always uncertainty about whether the results were comprehensive enough.
   ‍

With Innovation Gateway’s adoption of NLPatent, Kelsey was able to conduct high-quality patentability searches without relying on Boolean logic expertise.

Instead of spending hours crafting keyword logic, she could copy-and-paste invention disclosures directly into NLPatent using Natural Language Search, producing strong results even when she wasn’t confident she knew the “right” technical terminology.

From there, NLPatent’s Relevance Analysis and Ask NLPatent features helped her rapidly interpret the results, highlighting similarities and differences between the disclosure and each reference, acting as a built-in support layer when assessing unfamiliar inventions and terms of art.

As a result, searches that previously required budgeting a full day now typically take one hour or less, an ~90% reduction in time compared to previous methods.

The clarity of the relevance output also made it easier for Kelsey to communicate findings to inventors and internal stakeholders, supporting faster iteration and better alignment on patentability assessments.

NLPatent Turned a Post-Doc into the Office’s Internal Expert

Using NLPatent, Kelsey was able to demonstrate strong search capabilities early in her role. Today, she is a full-time staff member at Innovation Gateway and serves as the primary patent searcher, managing and training postdocs and supporting patentability review across the office.

For the broader TTO, the impact extends beyond time savings: it dramatically reduces the onboarding burden for new joiners.

Historically, a new hire may require:

• 15-20 hours learning how to interpret patent search results
  
• Plus additional time learning Boolean search construction
  

With NLPatent, Kelsey attests that new post-docs can become effective searchers within a few hours of use. From the TTO perspective, this increases productivity among new hires sooner and allows them to scale invention evaluations without relying on boolean search expertise.

As a result, NLPatent has become a core part of UGA’s patentability assessment workflow, enabling postdocs to contribute faster, improving confidence in results, and supporting stronger decision-making across a broad range of technology areas.