# patentnerd
PatentNerd: An AI Tool to Help Inventors Write Stronger Patents
Yuge Duan, Abigail Beckler, and Dennis Shasha
Computation, forthcoming (2026)
PatentNerd is an AI-powered tool that uses large language models to evaluate whether a patent’s claims are fully supported by its body, a common sticking point in patent litigation. While existing tools primarily focus on assessing novelty, PatentNerd emphasizes clarity and claim support, helping inventors strengthen the internal consistency of their applications.
The system also suggests more general claim language that remains faithful to the patent body, increasing robustness without sacrificing coverage. To evaluate its effectiveness, we analyze historical patent lawsuits to examine how PatentNerd’s analyses align with or anticipate judicial reasoning. The broader goal is to make the patent drafting process more reliable, defensible, and accessible for innovators.