USPTO Embraces Augmented Intelligence: A Look at Current AI Tools
The USPTO Inventor Hour on AI discussed AI tools like MLTD, Similarity Search, DocCode automation, and more
The United States Patent and Trademark Office (USPTO) is actively integrating artificial intelligence to enhance its operations, aligning with national directives to foster American leadership in AI. In a recent "USPTO Hour" on the topic, agency leaders discussed their progress, as well as some of their future plans.
Chief Information Officer (CIO) Jamie Holcombe highlighted the initiative's connection to Executive Order 14179, which aims to "sustain and enhance America's global AI dominance." With a clear mission to award patents, register trademarks, and lead the global IP ecosystem, the USPTO is strategically deploying AI to improve efficiency and quality.
Holcombe described the agency's approach as one of "augmented intelligence," emphasizing that the goal is to supplement the work of expert human examiners, not replace them. This frees up examiners from repetitive clerical tasks, allowing more time for thoughtful analysis. He noted that success has been found by carefully matching the right technology to the right task, a "right hammer for the right nail" philosophy.
This journey has underscored the critical importance of high-quality, curated data, as using incorrect data can lead to what Holcombe describes as "learning disabled" AI or "hallucinations." To this end, the USPTO has made its vast repository of 9 petabytes of patent data available to the public (and tool developers) through its Open Data Portal.
Examiners, practitioners, inventors and others dealing with IP should be wary of uploading or inputting potentially sensitive data into cloud tools or AI platforms due to the risk of premature public disclosure or otherwise exposing confidential information. Consult counsel.
Current AI Tools Enhancing Patent Examination
Acting Deputy Commissioner for Patents Greg Vidovich, another speaker at the event, detailed several AI-powered tools already yielding measurable results within the USPTO's Patents End to End (PE2E) search platform.
More Like This Document (MLTD): This feature assists examiners by identifying similar documents to one they are currently reviewing. With a simple click, the tool leverages AI to find related prior art from a database of over 120 million documents. This tool has seen significant adoption, with examiners using it nearly 850,000 times between March 2024 and February 2025.
Similarity Search: A more advanced tool, Similarity Search, ingests the full text of a patent application, including any amendments, and automatically searches for similar documents. The results are ranked for the examiner's review. The tool has been used for over 125,000 applications and has led to an increase in the citation of foreign prior art, enhancing examination quality. Notably, in over 30% of cases, at least one document found through Similarity Search is cited by the examiner on a PTO-892 form.
DocCode Quality Control: Shifting from search to process automation, the DocCode Quality Control tool is an in-house automated system designed to review the document codes applied by applicants to incoming filings. Correct document codes are essential for the proper routing and processing of the 18 million documents the USPTO receives annually. This tool, which entered full production in June 2025, reviews the applicant-applied code and either confirms it or suggests a correction, improving efficiency and reducing downstream errors.
Cooperative Patent Classification (CPC) Autoclassification: The USPTO is developing an AI tool to automate the application of CPC symbols to new patent applications. With over 250,000 CPC symbols, manual classification is a complex task. The in-house AI tool will apply CPC symbols to applications, helping to streamline the examination process and improve searchability for both examiners and the public.
DesignVision: Specifically for design patents, the DesignVision tool is intended to be a supplemental image-based search tool for design examiners. It takes the drawings from a design patent application and quickly searches for visually similar published designs. The goal is to launch this tool in (late) 2025.
The Next Frontier: Generative AI with SCOUT
Looking toward the future, Deputy Chief Information Officer Deborah Stephens presented on the agency's generative AI efforts, centered on its "SCOUT" (Searching, Consolidating, Outlining, and Understanding Tool) web application.
The webinar indicated that SCOUT was developed in a secure, controlled AI lab, and provides USPTO staff with access to state-of-the-art generative AI models. The platform began its alpha phase in March 2025 and now has over 200 users, with a beta release planned for later in the summer.
Current use cases being tested with SCOUT include:
35 U.S.C. § 112 Considerations: An "Antecedent Check" feature points out potential inconsistencies in patent claims.
Manual of Patent Examining Procedure (MPEP) Assistant: This function allows users to search the entire MPEP to quickly find answers to procedural questions.
Software Code Analysis: A "Developer Assistant" provides capabilities for software code analysis, refactoring (e.g., reorganizing, clarifying), and review.
The USPTO's current suite of AI tools shows a thoughtful and deliberate approach to integrating this powerful technology. By focusing on "augmented intelligence," the agency aims to empower its expert workforce, improve the quality and efficiency of examination, and ultimately strengthen the U.S. intellectual property system. The progress so far sets a strong foundation for future advancements.
Based on the webinar, the USPTO's current AI and automation efforts appear strategically focused on two fundamental pillars of the examination process: data integrity and prior art searching.
Tools like the DocCode Quality Control and CPC Autoclassification are foundational, ensuring that the immense volume of incoming data is categorized correctly from the outset. This creates a well-organized data environment, which is a prerequisite for effective searching.
Building on that foundation, tools such as "Similarity Search" and "More like this document" directly address the core examination task of finding relevant prior art for novelty and non-obviousness determinations under 35 U.S.C. § 102 and § 103. The clear objective is to augment the examiner's ability to conduct more comprehensive searches efficiently.
The SCOUT generative AI platform represents the next logical step in this progression. While the current tools focus on organizing data and finding references, SCOUT is poised to assist with the more nuanced analytical and drafting aspects of examination.
Its potential to help analyze applications for § 112 and § 101 issues and assist in drafting Office Actions points toward a future of more deeply integrated augmented intelligence for examiners. This layered approach suggests a deliberate strategy: first, get the data right; second, master the search; and third, leverage generative AI for higher-level analysis.
As that third higher level of analysis approaches, examiners and practitioners will all have to ensure there is still an independent, human analysis integrated into prosecution.
A future blog post will explore the details of the recent AI Request for Information (RFI), as presented by Director of the Office of Procurement Kristin Fuller, which outlines how the USPTO is looking to partner with the private sector to build upon these capabilities and further reduce patent pendency.
Disclaimer: This is provided for informational purposes only and does not constitute legal or financial advice. To the extent there are any opinions in this article, they are the author’s alone and do not represent the beliefs of his firm or clients. The strategies expressed are purely speculation based on publicly available information. The information expressed is subject to change at any time and should be checked for completeness, accuracy and current applicability. For advice, consult a suitably licensed attorney and/or patent professional.