USPTO’s Adoption of AI to Reduce Pendency in Patent and Trademark Applications

In a prior post back in January 2025, I discussed the potential for increased delays in patent and trademark applications due to numerous factors, including increased filings in both patent and trademark applications, hiring freezes for new examiners, possible reduction in the examiner workforce due to eliminated work-from-home positions, and across the board fee increases. In fact, in 2025, the United States Patent & Trademark Office (USPTO) saw major increases in pendency for both patent and trademark applications. The USPTO, aware of complaints from applicants and IP practitioners, has made concerted attempts to improve efficiency without compromising accuracy and quality. The Office has recently taken numerous steps to improve the application process for new and pending applications by introducing new AI tools developed by the agency to help streamline the process.

For years, initial processing of new trademark applications took little time. A first action could be expected within about 3 months from filing, and an application could mature to a registration well with one year from the filing date.  Recently, however, the time to receive a first action could take over 2-3 times as long, with full pendency for an application taking well over a year. Other critical steps for trademark applications, including processing of renewal petitions, reviewing Statement of Uses, and processing of assignments, have seen tremendous delays, frustrating stakeholders. For example, the average processing time for a Statement of Use in 2023 was about 9 days, well within the Trademark Office target of 15 days. By comparison, the processing time in the first quarter of 2026 was 157 days. Such an increase in processing time has mostly been attributed mostly to staffing reductions and the shifting of processing principles within the Office. Given growing frustration with such delays, the Office has been working to reduce processing time at all stages. A specific priority has been to get to new applications more quickly.

For example, to counter such trademark prosecution delays, the USPTO introduced a first-of-its-kind AI agent designed to handle some of the most time-consuming elements of trademark application pre-processing. The Trademark Classification Agentic Codification Tool, or “Class ACT”, utilizes AI technology to assign international classes to unclassified applications, determine appropriate design code to aesthetic features of new marks, and generate pseudo marks to expedite searching capabilities. USPTO Director John A. Squires recently commented that tasks such as classification and design search code processing could be reduced from a current backlog of about 5 months down to 5 minutes or even 5 seconds, which would greatly reduce the overall pendency time for new applications. Streamlining pre-processing of an application will allow examiners to substantively review applications more quickly and focus on what is needed the most – namely, providing appropriate guidance to applicants with skilled and timely examination. Director Squires noted that “[w]e’re providing specific, task-directed AI agents, which can efficiently tackle the toughest, most information intensive aspect of pre-examination – often the slowest step to achieve a complete application.” In effect, the Trademark Office is focused on providing immediate results with a higher level of accuracy. Based on the adoption of new tools like the Class ACT, the Trademark Office has been able to reduce the time for a first Office Action from 8 months last year to about 4.5 months at the beginning of 2026.  Similarly, the overall pendency for a trademark application has been reduced from 14.5 months to about 10.5 months.

Similar efforts have been made to reduce the pendency time for patent applications while increasing efficiency and enhancing accuracy. On April 10, 2026, the USPTO announced a significant milestone that had been reached in the first week of April. For the first time in nearly a decade, the cumulative output of the Office over a fiscal year was below the cumulative new filings over the same time period. Additionally, the number of unexamined patent applications dropped to its lowest levels in over 2 years. The Office has estimated that a one-week reduction in pendency for a patent application can increase value to the applicant by about $35,000. So, any reductions in the patent process will provide immediate economic benefits. As a result, the USPTO has introduced several new robust AI tools aimed to assist examiners in application review without compromising the quality of the review. These tools include products designed to accelerate prior art searching, automate invention classification, and streamline internal office communications. For example, the Patent Office has developed an image-based design search engine to facilitate prior art searches for design patent applications. Similarly, Patent Office has introduced SimSearch, a natural language search engine to assist examiners in searching U.S. and foreign patent databases, as well as ASAP!, the Artificial Intelligence Search Automated Pilot Program, designed to improve examination quality. The Patent Office has also developed SCOUT, the Searching, Consolidating, Outlining and Understanding Tool, which provides an internal generative AI chatbot to aid examiners in all patent prosecution tasks. These tools will be important for efforts to continue to chip away at the massive backlogs of application awaiting action from the Office.

Time will tell if these new AI tools will continue to reduce pendency of new applications. It is certainly a positive development to know that the USPTO is making concerted efforts to address the delays and provide applicants and IP practitioners with more timely feedback.