Everyone is focusing on the headline:
5 months to 5 minutes.
And that is impressive.
But the real significance of the USPTO’s new Class ACT tool is not that it makes trademark preprocessing faster. It is that it reveals something bigger about the future of trademark practice:
The USPTO’s March 19, 2026 announcement of Class ACT marks one of the clearest examples of useful, workflow-specific AI in trademark practice. The agency says the new Trademark Classification Agentic Codification Tool can immediately assign international classes, design search codes, and pseudo marks to unclassified applications, cutting classification preparation from about five months to five minutes, while the output is still reviewed by humans.
That matters especially for figurative marks.
Unlike plain word marks, figurative marks are harder to search because the relevant information is split across different layers: the wording, the image elements, the goods/services, and sometimes the way the wording sounds or is normalized for search. The USPTO’s field-tag guidance shows that trademark searching relies on separate searchable fields, including international class (IC), design code (DC), and a combined field that includes the word, pseudo mark, translation, and transliteration data.
So the real value of Class ACT is not that it “decides” the case. It is that it helps the USPTO convert a newly filed mark into searchable trademark data much earlier in the process.
Imagine an applicant files a logo for LION PEAK. The logo contains:
the wording LION PEAK
a mountain drawing
a star above the mountain
A typical application would include the image, the goods/services, and a mark description. From there, the office needs to make that application searchable.
With a figurative mark like this, Class ACT is designed to help generate three key pieces of metadata.
The class depends on the goods/services claimed in the application.
If the applicant is selling backpacks, the application might fall into one class; if it covers clothing, it might fall into another. The point is that the class tells the USPTO where the application fits in the trademark system and helps narrow examination and searching. The USPTO specifically says Class ACT can assign international classes to unclassified applications.
This is where figurative marks become especially interesting.
Because the mark includes visual elements, the USPTO needs searchable coding for those design components. In a logo like LION PEAK, the office may want the record to be searchable by elements such as:
mountain imagery
star imagery
other visual elements appearing in the drawing
The USPTO says Class ACT can assign design search codes, which historically were added by USPTO staff to make these applications easier to find.
That is a major operational gain. A figurative mark is not always easy to find by wording alone. If the image content is coded quickly, the examiner can begin searching visually similar marks much sooner.
Pseudo marks help normalize the wording of a mark for search purposes. The USPTO’s field-tag guidance confirms that search fields combine the word, pseudo mark, translation, and transliteration data.
In the LION PEAK example, the pseudo mark could simply preserve the readable wording:
Pseudo mark: LION PEAK
If the logo used unusual spelling, symbols, or stylized wording, pseudo-mark treatment would be even more important because it helps the system retrieve records that might not be captured by a literal text search alone. That is exactly why the USPTO identifies pseudo marks as one of the searchable elements Class ACT can generate.
Before tools like this, the office had to wait for people to code applications so they could be searched properly. The USPTO says this preprocessing delay had grown to several months and affected examination and customers. Now, according to the agency, AI can provide that information immediately at a high level of accuracy, while still leaving human review in place.
That helps the USPTO in three practical ways.
First, it reduces backlog at one of the most information-heavy points in the workflow. The press release expressly frames classification and design coding as one of the slowest parts of pre-examination.
Second, it improves early searchability. Figurative marks with logos, designs, and unconventional spelling are precisely the kinds of applications the USPTO says can be difficult to search. Class ACT targets that problem directly.
Third, it frees examining attorneys and staff to spend more time on substantive judgment rather than on repetitive metadata preparation. The USPTO says this lets employees focus their experienced judgment and reason on substantive issues in examination.
For practitioners, Class ACT is a sign that trademark administration is becoming more data-structured and less dependent on slow manual preprocessing.
That means two things.
One, the office may be able to get to a searchable record much faster. Two, the quality of the filing itself becomes more exposed. If the office is no longer delayed by coding and class-prep bottlenecks, then weak goods/services drafting, poor mark descriptions, or unclear filing strategy stand out sooner.
So even though Class ACT is an internal USPTO improvement, it has an external effect on practice: it increases the value of precision at filing.
The figure above is a simplified visual of how this works:
the applicant files a figurative mark
Class ACT helps generate searchable data
the examiner can then search by class, design elements, and wording-related fields
One note: the image is illustrative, not an official USPTO diagram, and some text inside the generated graphic is imperfect. The concept, however, reflects the workflow described by the USPTO.
Class ACT helps the USPTO by turning figurative marks into searchable records faster. For a logo mark, that means earlier assignment of:
international class
design search codes
pseudo-mark data
That is not a replacement for legal analysis. It is something more practical: a way to reduce clerical drag and make the search-and-examination system work sooner and better. And for figurative marks, that is exactly where AI can deliver real value.