The US Patent and Trademark Office has already invested heavily in artificial intelligence capabilities. The employee will access several tools to review documents and reduce the burden on the examiner of administrative and administrative tasks.
USPTO wants to accelerate the use of AI tools through new requests for information.
At RFI, USPTO says it is looking for technologies that can improve the efficiency of its patent and trademark inspection processes. Agency highlighted by RFI is new code or ideas for improving its surroundings, developing/using Robotic Process Automation (RPA) bots, and improving current processes.
Greg Vidvich, acting vice-chairman of patents, said vendors should submit feedback through the RFI to help the agency expand its current capabilities.
“Essentially [the RFI] Equivalent to a summary of the use of AI/machine learning to complete a comprehensive search report. In the second bullet, this would utilize artificial intelligence to create office actions to properly determine formal objections, such as abstract specifications claims, numbers and rejections,” he said on June 17th in the USPTO Industry Day. Therefore, whatever the examiner can provide or provide input can be created or drafted an office action. ”
RFI asks 7 questions
One interesting thing about RFI is that USPTO is looking for low-cost or no-cost services.
This was one of seven questions USPTO asked for feedback.
“Discuss your willingness to provide a solution in exchange for low, no cost or non-financial considerations. If low-cost solutions or non-financial benefits are interesting, explain how low-cost and non-financial benefits are interested in your/your organization.”
Another question asked the vendor to “explain innovative tools and solutions, especially AI-based tools or solutions. This was developed to help with comprehensive search and analysis of previous art.”
USPTO posted its RFI on sam.gov on June 4th. The vendor submitted questions until June 10th, and the USPTO has extended the deadline for answers until July 1st.
Kristen Fuller, director of the USPTO Procurement Bureau, said that once the RFI is closed, the agency will reach out to respondents and set up a one-on-one meeting to learn about their abilities and how they responded.
Regarding the possibility of solicitation, Fuller said he was not ready to talk about the USPTO's issuance plan.
However, given what USPTO has done so far and the possibilities that it may be seen in AI capabilities, it is clear that agents want to continue accessing these tools.
USPTO Chief Information Officer Jamie Holcombe said the agency's AI journey began four years ago.
“We first adopted a stock of nine petabytes of detailed patent data, which anyone can download from the open data portal at data.uspto.gov and see for themselves the insights hidden in this treasury of intellectual property data,” Holcomb said. “We created a machine learning pilot, which led to production applications in the field of classification and citation, and has had previous art searches and fraud detection. That all AI needs to know the data structures, data elements, data flows, and most importantly, data security. Computing requires sources that are curated data, the creation of golden datasets, and where to start. This is because training with incorrect or inconsistent data often results in applications being labeled “Learning Disabled.” Most others call them hallucinations with inaccurate consequences. ”
Holcombe said that despite the USPTO having several ongoing programs using AI, it believes USPTO is at the beginning of its journey to using AI.
Vidovich said one tool is an AI assist feature for examiners that helps them find similar documents more effectively while performing searches.
“This allows examiners to find documents related to the documents they are reviewing as part of their existing search. This tool is part of the patent end-to-end (PE2E) search tool. It is an internal search tool built on a modern web-based platform in the cloud,” he said. “This document (MLTD) searches for US patents, pre-publications and foreign patents, and the tool runs more than 120 million fully indexed documents, which are intended to complement the examiner.
Existing AI Features
Since USPTO launched the tool in March 2024, Vidovich said that examiners have used this nearly 850,000 times in their offices, with the adoption rate increasing in 2025.
The second tool, called Doc Code Quality Control, is a new automation tool for viewing all incoming documents from applicants through the patent center.
“What we did inside is we have built automation tools in our cars. It is designed to check all incoming documents from applicants through our patent centre,” Vidovich says. “This tool essentially reviews the applicant's documentation and considers the document code applied by the applicant. And if it's correct, it suggests that it's an internal alternative to make that change. Why is this important? For this, we plan to reduce contract dependencies and provide cost savings to our agents.”
In addition to RFI and existing tools, USPTO has several AI tools in the early testing stage where you plan to expand. Many of the tools use generator AI.
Deborah Stephens, the USPTO's Deputy CIO CIO, said agencies can make early recruits accessible to AI labs, and teams can build tests, validate use cases and assess their impact on the business.
According to Stephens, one of the USPTO's test Genai tools is called scouts: search, integration, overview, and understanding.
“We leverage scouts to help develop code, detect inappropriate filings, and assist in cybersecurity threat detection and compliance, as well as monitoring and correlation,” she said. “Like many people, we saw the global rollout of AI and its potential. In May 2023, a small federal team began using frontier AI models with safe and appropriate guardrails, which were only available to this very technical internal user. Fast forward to March 2025, over 40 users have won very specific use cases and will release a larger user-based beta this summer.
Stephens believes Scout is making a difference for employees because he believes that its capabilities are valuable and proves that it makes a difference for employees.
Copyright©2025 Federal News Network. Unauthorized reproduction is prohibited. This website is not intended for users within the European Economic Area.
