USPTO Issues 2 Memoranda Regarding SMED Term to Patent Examinatio

The United States Patent and Trademark Office has given patent applicants a fresh reason to dust off a tool that was always sitting in the prosecution toolbox but often treated like the odd wrench nobody wanted to use. On December 4, 2025, the USPTO issued two memoranda addressing Subject Matter Eligibility Declarations, commonly shortened to SMEDs. One memorandum was written for the Patent Examining Corps. The other was written for patent applicants and practitioners. Together, they clarify how factual evidence may be used during patent examination to respond to subject matter eligibility rejections under 35 U.S.C. § 101.

That may sound like a sentence only a patent attorney could love, but the practical effect is important. For years, inventors in software, artificial intelligence, diagnostics, fintech, networking, data processing, and other high-tech fields have faced a familiar prosecution headache: an examiner says the claims are directed to an abstract idea, a law of nature, or another judicial exception. The applicant responds, “No, this is a real technical improvement.” The examiner replies, “Still abstract.” Everyone sighs. The file history grows. Somewhere, a coffee machine loses the will to live.

The SMED guidance does not change the law. It does not rewrite Alice, Mayo, Myriad, or any of the major court decisions that shaped modern eligibility analysis. What it does is clarify that applicants may submit evidence under 37 C.F.R. § 1.132 to help show why a claimed invention is patent-eligible. More importantly, it reminds examiners that properly submitted SMED evidence must be considered as part of the record and weighed under the preponderance-of-the-evidence standard.

What Is a SMED?

A Subject Matter Eligibility Declaration is a declaration submitted under Rule 132 that focuses specifically on a subject matter eligibility rejection. In plain English, it is a structured evidentiary statement used to explain facts relevant to whether the claimed invention is the kind of invention that can be patented under § 101.

A SMED might explain how a claimed machine-learning system improves computational performance, how a network-monitoring process cannot practically be performed in the human mind, or how a medical diagnostic invention applies a natural relationship in a practical technological way. It is not meant to be a dramatic courtroom monologue. It is meant to be factual, focused, and tied directly to the claims.

The USPTO’s guidance emphasizes that a SMED should have a clear nexus to the claimed invention. That word “nexus” is doing a lot of work. It means the evidence cannot float around the application like a motivational poster. The declaration must connect the facts to the actual claim language and the disclosure in the specification.

Why the USPTO Issued Two Memoranda

The first memorandum, directed to patent examiners, explains how examiners should treat SMEDs during examination. It reminds the examining corps that Rule 132 declarations are already part of existing USPTO practice and that applicants may use them to submit factual evidence relevant to eligibility. The memo also makes clear that SMEDs are optional. No applicant is required to file one. But when a properly submitted SMED appears in the record, the examiner cannot simply wave it away like an inconvenient parking ticket.

The second memorandum, directed to applicants and practitioners, focuses on best practices. Its headline message is simple: submit the SMED as a separate document when addressing subject matter eligibility. Do not blend eligibility testimony with obviousness arguments, written-description explanations, novelty evidence, or other issues unless there is a very good reason. Combining everything into one giant declaration may feel efficient, but it can create a prosecution smoothie where nobody can tell which ingredient is supposed to prove what.

The Legal Backdrop: Section 101 and the “Something More” Problem

Subject matter eligibility under 35 U.S.C. § 101 is the gatekeeper question in U.S. patent law. Before the USPTO decides whether an invention is new, non-obvious, or adequately described, it asks whether the claimed subject matter falls within patent-eligible categories and avoids judicial exceptions such as abstract ideas, natural phenomena, and laws of nature.

The modern difficulty comes from applying the Alice/Mayo framework. Under that framework, claims may be rejected if they are directed to a judicial exception and do not include significantly more than that exception. The phrase “significantly more” has generated enough confusion to power a small city. In some technologies, especially software and AI, applicants often argue that their claims improve computer functionality or integrate an abstract idea into a practical application. Examiners may disagree. The SMED guidance addresses this exact zone of friction.

The USPTO’s memoranda build on a broader subject matter eligibility conversation, including the agency’s 2024 AI eligibility update, later examiner reminders, and the precedential Ex parte Desjardins decision. In Desjardins, the USPTO emphasized that improvements in computational performance, learning, storage, data sets, and data structures can constitute patent-eligible technological advancements when properly reflected in the claims and specification.

What the Examiner Memorandum Says

The examiner-focused memorandum explains that SMEDs may help establish patent eligibility when the technical improvement is not readily apparent from the examiner’s reading of the application. It points examiners to existing USPTO guidance, including MPEP sections dealing with declarations and subject matter eligibility. The memo does not create a new filing procedure. Instead, it clarifies how an existing evidentiary path may be used in the § 101 context.

The memorandum also warns that conclusory opinions are not enough. A declaration that merely says, “This invention is totally eligible, trust me,” is unlikely to move the needle. A stronger SMED explains the factual basis for the conclusion. For example, it may include technical test data, comparative performance results, expert explanation, or evidence about how a person of ordinary skill in the art would understand the specification at the time of filing.

The examiner memo also makes the USPTO’s expectation clear: examiners must carefully consider the applicant’s arguments and evidence. If a SMED is persuasive, the examiner should explain why the eligibility rejection is withdrawn. If the SMED is not persuasive, the examiner should explain why the rejection is maintained. That requirement matters because it improves the written record and gives applicants a clearer path for response, interview, continuation strategy, or appeal.

What the Applicant Memorandum Says

The applicant-facing memorandum delivers a practical drafting lesson: keep the SMED separate from other declarations. A Rule 132 declaration can address different kinds of rejections, but the USPTO strongly encourages applicants to avoid mixing eligibility evidence with testimony on obviousness, secondary considerations, motivation to combine, enablement, written description, or novelty.

That separation helps everyone. The examiner can evaluate eligibility evidence without sorting through unrelated issues. The applicant creates a cleaner prosecution record. Future reviewers, including the Patent Trial and Appeal Board, can see exactly what evidence was offered for § 101 and why. In patent prosecution, clarity is not just good manners. It is a survival strategy.

The memorandum also states that SMEDs should not be used to supplement the original disclosure. That is a crucial limitation. A declaration may clarify facts, explain how a skilled artisan would understand the application, or provide objective evidence tied to the claims. It cannot add new technical content that should have been in the specification on the filing date. In other words, a SMED can polish the window, but it cannot build a new house behind it.

Examples of When a SMED May Help

Software and AI Inventions

For software and artificial intelligence applications, a SMED may help when the claims are rejected as abstract despite reciting a concrete technical improvement. Suppose an application claims a machine-learning model training technique that reduces storage requirements and prevents loss of prior learning. A declaration from a technical expert could explain how the claimed architecture changes the operation of the computer system rather than merely applying a business idea on a generic computer.

Network Security and Distributed Systems

In network security, an examiner may characterize claim steps as mental processes, such as detecting suspicious activity or analyzing information. A SMED could explain why the claimed distributed packet analysis across multiple machines cannot practically be performed in the human mind and why the architecture improves network monitoring. That evidence may directly rebut a factual assumption underlying the rejection.

Medical Diagnostics

For diagnostics, the eligibility problem often involves natural laws or correlations. A SMED might help show that the claims are not merely observing a natural relationship but applying it through a specific technological process, assay configuration, or treatment-related workflow. The declaration should stay close to the claims and specification because medical diagnostics live in a particularly sensitive § 101 neighborhood.

What a Strong SMED Should Include

A strong SMED begins with a clear identification of the rejection being addressed. It should then explain the relevant claim limitations, the parts of the specification that support the technical improvement, and the factual evidence showing why the invention is eligible. The declarant may be an inventor, engineer, scientist, technical employee, or independent expert with credible knowledge of the facts.

The most effective declarations are specific. They do not merely repeat attorney argument in affidavit form. They explain how the technology works, why the improvement matters, and how the claim language captures that improvement. If comparative data is available, it should be presented clearly. If industry literature supports the factual point, it can be referenced. If the invention improves latency, accuracy, memory usage, power consumption, network throughput, data structure efficiency, or system reliability, those improvements should be described in concrete terms.

Applicants should also cross-reference the SMED in the accompanying response. The response should identify the exact declaration paragraphs that rebut specific examiner findings. That way, the examiner does not have to play hide-and-seek with the evidence. Patent prosecution already has enough puzzles.

What a SMED Should Not Do

A SMED should not introduce new matter. It should not rewrite the invention after the fact. It should not make broad legal conclusions without factual support. It should not argue that the invention is eligible simply because the applicant really, really wants a patent. Hope is not evidence, even when printed on firm letterhead.

A SMED also should not blur § 101 with § 103. Obviousness declarations often focus on secondary considerations, commercial success, long-felt need, failure of others, or motivation to combine. Eligibility declarations focus on whether the claimed invention is directed to patent-eligible subject matter and whether it integrates any judicial exception into a practical application or recites significantly more. Those are different questions. The USPTO’s guidance encourages applicants to keep them in separate lanes.

How the April 2026 Update Fits In

In April 2026, the USPTO issued an updated best-practices memorandum that superseded the December 2025 applicant-facing memorandum. The update reflected feedback from applicants and practitioners during early implementation of the voluntary SMED program. It also signaled that the USPTO views SMED guidance as a living practitioner resource, especially as AI-related patent examination continues to evolve.

This matters for practitioners because it means SMED practice is not frozen in time. Applicants should monitor USPTO updates, examiner training materials, MPEP changes, and PTAB decisions. A SMED strategy that works well today may need refinement as the USPTO gathers more experience with these declarations. Patent law is many things, but it is rarely a still pond.

Practical Strategy for Patent Applicants

Applicants should consider a SMED early when a § 101 rejection appears to rest on a factual misunderstanding. If the examiner overlooks a technical improvement, treats a computer-implemented invention as generic automation, or assumes a process can be performed mentally when it cannot, a SMED may be useful. Filing it with the first substantive response may create a stronger record than waiting until prosecution has already become a procedural tug-of-war.

An examiner interview can also help. Before preparing a declaration, counsel may ask what facts the examiner considers missing or unpersuasive. That conversation can shape a more targeted SMED. The goal is not to bury the examiner under paper. The goal is to answer the factual question that is actually blocking allowance.

Applicants should also think about litigation consequences. A declaration becomes part of the prosecution record. Statements made to secure allowance may later be studied by challengers, accused infringers, and courts. That does not mean applicants should avoid SMEDs. It means they should draft them carefully, accurately, and with future enforcement in mind.

Experiences and Lessons from SMED-Focused Patent Examination

In practical patent prosecution, the most useful lesson from the USPTO’s SMED memoranda is that eligibility arguments become stronger when they sound less like abstract legal debate and more like engineering reality. Many applicants have experienced the frustration of explaining a technical invention only to receive an Office action that compresses the innovation into a high-level concept. A cybersecurity invention becomes “collecting and analyzing data.” A machine-learning training method becomes “mathematical processing.” A diagnostic workflow becomes “observing a natural relationship.” Once the claim is described at that altitude, oxygen gets thin fast.

A well-prepared SMED can bring the discussion back down to the ground. Instead of arguing only that the examiner applied the eligibility test incorrectly, the applicant can submit evidence showing what the technology actually does. For example, an inventor might explain that the claimed distributed architecture changes where processing occurs, reduces bottlenecks, and allows real-time detection that was not possible with conventional centralized systems. An independent expert might explain how a claimed data structure improves memory retrieval in a way a skilled artisan would recognize as a computer-functionality improvement. Those facts can give the examiner a more complete record.

Another practical experience is that the best SMEDs are built from good specifications. If the original application describes the technical problem, the prior limitations, and the measurable improvement, the declaration has something solid to clarify. If the specification is vague, the SMED has less room to operate because it cannot add new matter. This is why the guidance should influence drafting behavior even before examination begins. Patent applications in AI, software, diagnostics, and data systems should describe not only what the invention does but how it improves technology in a concrete way.

There is also a workflow lesson for law firms and in-house teams. SMED preparation should not be treated as a last-minute patch. It often requires coordination among patent counsel, inventors, technical experts, and sometimes litigation counsel. The declaration should be accurate, technically grounded, and consistent with the application. The response should cite the declaration clearly. The prosecution team should avoid exaggerated claims that feel helpful in the moment but create problems later. In patent practice, the file history has a long memory and a surprisingly good search function.

Finally, the SMED memoranda encourage a more disciplined conversation between applicants and examiners. They do not guarantee allowance. They do not make every software or AI claim eligible. But they give applicants a clearer way to place facts into the record and require examiners to address those facts. That is a meaningful improvement. Patent examination works best when the debate is not merely about labels like “abstract idea,” but about the claimed invention, the specification, the evidence, and the real technical contribution. If SMEDs help move more cases into that kind of focused analysis, they will have earned their place in the prosecution toolbox.

Conclusion

The USPTO’s two SMED memoranda mark an important development in patent examination, especially for applicants facing § 101 rejections in software, artificial intelligence, diagnostics, and other complex technologies. The guidance does not change the statutory standard or create a new procedure. Instead, it clarifies how applicants may use Rule 132 declarations to present factual evidence and how examiners should evaluate that evidence.

The central message is practical: eligibility can involve factual questions, and factual questions deserve factual evidence. A SMED should be separate, focused, timely, objective, and tied closely to the claims and specification. Used properly, it can clarify the record, sharpen examiner review, and help applicants explain why their inventions are more than abstract ideas dressed up in patent clothing.