With the boom in artificial intelligence (AI) and machine learning (ML) innovations, it has felt like examiners have been on a mission to issue 101 rejections and not listen to (what I think is) common sense. But as we know, the pendulum tends to swing back. 

On Aug. 4, 2025, the United States Patent and Trademark Office (USPTO) issued a memo reminding examiners how to apply 101 in the AI/ML space. The gist? Examiners should look at claims as a whole, distinguish between merely involving an exception versus reciting one, and only issue a 101 rejection if they meet the burden of proof — meaning it’s more likely than not the claims are patent ineligible.

This memo turned out to be a game-changer for a couple of stubborn cases that had rounds of Requests for Continued Examination (RCE) and extensive independent claims. 

So what works?

• Structure your arguments clearly. Walk through the memo’s logic as your framework for arguing.
• Emphasize the difference between “reciting” versus “involving” an exception. Amend where necessary to shift into the ‘involve’ framework. Otherwise, explain why your claims as a whole merely involve an exception and thus the 101 inquiry should stop there. For example, if the claims include techniques that merely relate to known mathematical concepts, they do NOT recite mathematical concepts.
• Clearly explain technological improvements. Lean on your spec and what a person of ordinary skill in the art would understand as the technical problem being solved. Focus on improvements to machine learning training and/or refinement techniques. Also, consider how your processing or model use is more computationally efficient.
• Analyze the claims as a whole with every argument. Remind the examiner that they cannot look at any feature in isolation as their basis for rejection.  
• Push back on mental process overreach. If a step can’t be done in the human mind (e.g., data processing, model training, real-time determinations from robust dataset ingestion and processing), call it out loud and clear. Highlight the obvious arguments too: the human mind does not have the capacity of CPUs, memory, cache — you name it — to process data at lightning speed.
• Call out the examiner on the burden of proof. The memo indicates that the examiner should only make a rejection when it is more likely than not (more than 50%) that the claim is ineligible. Conclude your arguments with an indication that the examiner has not met this burden.

If you're stuck in a 101 loop, don’t lose hope. The memo gives us a solid framework — and sometimes, a well-structured argument might be all it takes to tip the scales in your favor.