The Office's opt-in Artificial Intelligence Search Automated Pilot (ASAP!) Program, announced in the October 8, 2025 Federal Register notice (90 FR 48161), used an application's CPC designation, specification, claims, and abstract to surface up to ten potentially relevant prior-art documents before formal examination. The pilot closed on June 1, 2026. Director Squires has described the program as the first of several planned AI initiatives, indicating that AI-assisted search capabilities are likely to expand further.
In a recent Bloomberg Law analysis, Michael Anderson of Marshall, Gerstein & Borun LLP distinguishes two categories of tools the USPTO is developing: examiner-assistant tools that surface higher-quality prior art for human review, and agentic tools that can autonomously search databases, iteratively evaluate results, and potentially draft initial office actions. Both approaches are expected to identify more on-point prior art, more quickly, and with less reliance on traditional keyword matching.
What Patent Applicants Should Consider
- Improved AI search tools are likely to surface references that conventional searches miss, increasing the importance of thorough technical disclosure.
- More pertinent prior art can lead to additional prosecution rounds, more narrowing amendments, and a higher likelihood that weaker applications will be abandoned. Gains in examination speed per application may not translate into shorter overall pendency.
- Software and computer-related inventions may be particularly affected, as they are often filed earlier in the development cycle with less mature technical disclosure than mechanical or biological inventions.
The practical response is to file robust, detailed specifications that explain not only what an invention achieves but how it achieves it, with engineer-level technical specificity. Applications closer to actual reduction to practice will generally have the disclosure depth needed to support meaningful claim amendments if particularly relevant prior art is later identified.
This emphasis on specification quality aligns with the principles reflected in Ex parte Desjardins, Recentive Analytics, and the USPTO's SMED (Subject Matter Eligibility Declaration) guidance. Whether the issue arises under § 101 or §§ 102/103, the strength of the written description remains central to successful AI patent prosecution as search tools continue to improve.
Read more from Bloomberg Law: https://lnkd.in/gAqvNN6z