SMEDs After the USPTO’s 2026 Guidance: Rule 132 Declarations, Corroboration, and the Need for a Court-Ready § 101 Record

Executive Summary: This article argues that Subject Matter Eligibility Declarations (“SMEDs”) can be valuable prosecution tools for building a factual record in response to § 101 rejections, but only if they are used with discipline. A SMED should not ask an inventor or expert to declare that claims are patent eligible; rather, it should provide claim-tethered, corroborated technical facts showing how a person of ordinary skill in the art would understand the specification and claims as reciting a concrete technological improvement. The article explains that USPTO SMED guidance may help applicants frame eligibility evidence during examination, but it does not alter Federal Circuit law or relax the requirement that the claims themselves recite the improvement. Using two Swoop declarations as case studies, the article shows how effective SMEDs can reframe email-based commerce claims as technical security architectures: one involving a UUID-linked, expiration-bound authentication control packet for returned SMTP/mailto order emails, and the other involving stateful, cryptographically verifiable token security for SMTP-based transaction messages. The central lesson is that SMEDs are not eligibility shortcuts; they are record-building tools whose persuasive force depends on nexus, corroboration, specification support, and a claim-centered technological narrative.

I.             Introduction

Patent eligibility disputes under 35 U.S.C. § 101 are formally legal questions, but the hard cases often turn on technical facts. Since Alice, courts and examiners have asked whether claims are directed to a judicial exception and, if so, whether the claim elements, individually or as an ordered combination, add enough to transform the claim into patent-eligible subject matter. Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 217–18 (2014). That framework is legal in structure, but it often depends on underlying factual propositions: whether the claim changes computer functionality, whether an ordered combination was well-understood, routine, and conventional, whether a limitation can practically be performed in the human mind, and whether the specification would have conveyed a technological improvement to a person of ordinary skill in the art.

That factual dimension is the proper domain of a Subject Matter Eligibility Declaration, or SMED. A SMED is a declaration under 37 C.F.R. § 1.132 submitted to address a subject matter eligibility rejection. Properly used, it does not ask the declarant to opine that the claims are patent eligible. It asks the declarant to provide technical facts that support the eligibility argument.

The USPTO’s December 4, 2025 examiner-facing memorandum states that a SMED may show how a person of ordinary skill in the art would interpret a specification that describes a technological improvement, while also emphasizing that evidence must have a nexus to the invention as claimed, objective evidence should be supported by actual proof, and legal conclusions on eligibility are not entitled to weight. Memorandum from John A. Squires, Under Sec’y of Com. for Intell. Prop. & Dir. of the U.S. Patent & Trademark Office, to Patent Examining Corps, Subject Matter Eligibility Declarations  (Dec. 4, 2025) (“Examiner SMED Memo”).

The April 30, 2026 practitioner memorandum makes the same point from the applicant side. It treats SMEDs as evidence specifically relevant to § 101, emphasizes nexus, warns that SMEDs must not improperly supplement the specification, and states the most important limiting principle: “Claims always define the invention.” Memorandum from John A. Squires, Under Sec’y of Com. for Intell. Prop. & Dir. of the U.S. Patent & Trademark Office, to All Patent Applicants and Patent Practitioners, Best Practices for Submission of Rule 132 Subject Matter Eligibility Declarations (SMEDs) 3 (Apr. 30, 2026) (“Apr. 2026 SMED Memo”).

II.          SMEDs Are Record-Building Tools, Not Eligibility Shortcuts

The most important practical point is also the easiest to overstate. A SMED is not a substitute for claim drafting. It cannot import an unclaimed embodiment into the claims, cannot cure missing disclosure, and cannot convert a result-oriented claim into a technological implementation. That limitation is not merely a matter of USPTO preference; it reflects the Federal Circuit’s claim-centric approach to § 101.

Claims that merely use a technological environment do not become eligible simply because the surrounding field is technical. ChargePoint, Inc. v. SemaConnect, Inc., 920 F.3d 759, 766–74 (Fed. Cir. 2019). Claims that make a computer process easier without reciting a technological means of improvement remain vulnerable. Simio, LLC v. FlexSim Software Prods., Inc., 983 F.3d 1353, 1360–65 (Fed. Cir. 2020). Even in a mechanical manufacturing context, a claim cannot simply recite a desired result without claiming the means that apply the law of nature in a specific way. Am. Axle & Mfg., Inc. v. Neapco Holdings LLC, 967 F.3d 1285, 1292–1301 (Fed. Cir. 2020).

That claim-centric limitation is reinforced by the hierarchy of authority governing § 101. USPTO guidance can be important in prosecution, but it does not alter the substantive law applied by the Federal Circuit or district courts. In In re Rudy, the Federal Circuit stated that USPTO eligibility guidance “is not, itself, the law of patent eligibility,” does not carry the force of law, and does not bind the court’s eligibility analysis. 956 F.3d 1379, 1382–83 (Fed. Cir. 2020). The court has made the same basic point in other eligibility cases. See In re Smith, 815 F.3d 816, 819 (Fed. Cir. 2016); Cleveland Clinic Found. v. True Health Diagnostics LLC, 760 F. App’x 1013, 1020 (Fed. Cir. 2019); In re Abel, 838 F. App’x 558, 560 (Fed. Cir. 2021). Nor are courts bound by PTAB decisions merely because they have precedential status within the agency. ENCO Sys., Inc. v. DaVincia, LLC, 845 F. App’x 953, 957–58 (Fed. Cir. 2021).

Thus, the practical role of USPTO guidance is evidentiary and procedural. It may help an applicant frame and develop the prosecution record. It may also give examiners a reason to engage technical evidence more directly. But it does not create a different eligibility standard. The gap between Office receptivity and litigation durability remains real.

That gap should shape SMED strategy. A SMED should be drafted to satisfy the examiner’s evidentiary expectations under Rule 132 and USPTO guidance, but it should also create a record that can survive later judicial scrutiny. The declaration should therefore identify facts that matter under Federal Circuit doctrine: how the claim changes computer or network operation; why the ordered combination was not well-understood, routine, and conventional; why the alleged improvement is reflected in the claim language; and how the specification would have conveyed that improvement to a person of ordinary skill in the art. See Berkheimer v. HP Inc., 881 F.3d 1360, 1368–70 (Fed. Cir. 2018); Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335–39 (Fed. Cir. 2016); BASCOM Glob. Internet Servs., Inc. v. AT&T Mobility LLC, 827 F.3d 1341, 1349–52 (Fed. Cir. 2016).

The same caution applies to Ex parte Desjardins. Desjardins is useful because it illustrates a claim-centered path for eligibility where the claim reflects an improvement in the operation of an AI model itself. Ex parte Desjardins, Appeal No. 2024-000567, slip op. at 7–10 (P.T.A.B. Sept. 26, 2025) (Appeals Review Panel), designated precedential Nov. 4, 2025. But Desjardins is not a general safe harbor for software, AI, fintech, diagnostics, e-commerce, or other eligibility-sensitive technologies. Its usefulness depends on whether the pending claims recite the technological improvement, not merely whether the specification describes one or the applicant characterizes the field as advanced.

Recentive Analytics supplies the limiting principle. There, the Federal Circuit held that applying generic machine-learning techniques to event scheduling and network-map generation did not make the claims patent eligible where the claims did not improve machine-learning technology itself. Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025). The contrast between Desjardins and Recentive is instructive, though not perfectly symmetrical. Desjardins reflects an Office-side path for claims that improve model operation; Recentive reflects the judicial limit for claims that merely deploy generic AI in a new business or operational environment. An AI-related claim is stronger when it recites a specific improvement to model operation, training architecture, storage, computational performance, or data structures. It remains vulnerable when it merely applies generic AI or software functionality to a new field of use.

The better view, then, is modest but useful: a SMED should not ask the Office to accept a different § 101 standard. It should create a disciplined factual record showing that the claims, as written and supported by the specification, recite a concrete technological improvement under the governing case law.

III.      The POSITA’s Role in § 101 Is Real but Derivative

The person of ordinary skill in the art plays a central role in patent law, but not the same role in every statutory inquiry. Section 103 expressly asks whether the claimed invention would have been obvious to a POSITA. 35 U.S.C. § 103. Section 112 often turns on what the skilled artisan would have understood from the disclosure. Claim construction asks how the claim language would have been understood in view of the intrinsic record. Phillips v. AWH Corp., 415 F.3d 1303, 1312–17 (Fed. Cir. 2005) (en banc).

Section 101 is different. The POSITA does not decide eligibility. Instead, the POSITA enters eligibility through factual predicates: conventionality, technological improvement, mental-process characterization, practical application, and claim meaning. That distinction is easy to blur in prosecution. A declaration that sounds like a legal brief may be tempting, but it is less useful than testimony that proves the technical facts on which the legal argument depends.

The Federal Circuit recognized this factual dimension in Berkheimer v. HP Inc., holding that whether a claim element or ordered combination is well-understood, routine, and conventional is a question of fact. 881 F.3d at 1368–70. Aatrix Software, Inc. v. Green Shades Software, Inc. likewise recognized that factual allegations concerning unconventionality or improved computer functionality can matter at the pleading stage. 882 F.3d 1121, 1128–30 (Fed. Cir. 2018).

Federal Circuit cases provide recognized eligibility pathways where technical facts matter. Claims directed to a self-referential database table can be eligible when focused on a specific improvement in computer functionality. Enfish, 822 F.3d at 1335–39. Claims using particular rules to automate animation can be eligible when they recite a specific improvement rather than merely invoking a result. McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1313–16 (Fed. Cir. 2016). An inventive concept may reside in a nonconventional and non-generic arrangement of known components. BASCOM, 827 F.3d at 1349–52. Network-monitoring and cardiac-monitoring claims can be eligible when they recite specific technological improvements. SRI Int’l, Inc. v. Cisco Sys., Inc., 930 F.3d 1295, 1303–04 (Fed. Cir. 2019); CardioNet, LLC v. InfoBionic, Inc., 955 F.3d 1358, 1368–71 (Fed. Cir. 2020).

A SMED should be built around those factual predicates. The declarant may explain what the specification would have conveyed to a POSITA, what the state of the art was, why a limitation cannot practically be performed mentally, why an ordered combination was not conventional, or how the claim changes system operation. The declarant should not merely announce that the claims are “not abstract” or “patent eligible.”

IV.       Corroboration Is the Difference Between a Useful SMED and a Vulnerable One

The USPTO guidance permits a SMED to be signed by any person with knowledge of the asserted facts, including an inventor, co-worker, independent expert, or other knowledgeable witness. Examiner SMED Memo 2. But admissibility and probative value are different things. A properly signed inventor declaration may be entered and considered, but it will be more persuasive if it is corroborated.

This analogy has limits. The classic corroboration cases are not SMED cases, and they do

not create a categorical rule that inventor-signed SMEDs are insufficient. But they reflect a durable evidentiary instinct in patent law: interested testimony about technically significant facts is stronger when supported by objective evidence.

In The Barbed Wire Patent, the Supreme Court warned about oral testimony concerning prior use, noting the dangers of faded memory, exaggeration, and self-interest. 143 U.S. 275, 284–85 (1892). In Eibel Process Co. v. Minnesota & Ontario Paper Co., the Court emphasized the value of contemporaneous records over later testimonial reconstruction. 261 U.S. 45, 60–61 (1923).

The Federal Circuit’s corroboration cases carry the same caution forward. In Price v. Symsek, inventor testimony alone was insufficient to prove conception, derivation, or priority, and corroboration was assessed under a rule of reason. 988 F.2d 1187, 1194–95 (Fed. Cir. 1993). Woodland Trust v. Flowertree Nursery, Inc. warned against reliance on uncorroborated oral testimony concerning invalidating prior use. 148 F.3d 1368, 1371–73 (Fed. Cir. 1998). Finnigan Corp. v. International Trade Commission reiterated the need for corroboration when testimony concerns prior invention or reduction to practice. 180 F.3d 1354, 1367–70 (Fed. Cir. 1999). Singh v. Brake stated the policy concern directly: interested witnesses may be tempted to remember or characterize events in a self-serving way. 222 F.3d 1362, 1367 (Fed. Cir. 2000).

The same evidentiary lesson applies to SMEDs. The issue is not whether inventor testimony is allowed. It is. The harder question is how much weight the testimony should receive. In EmeraChem Holdings, LLC v. Volkswagen Group of America, Inc., the Federal Circuit rejected unsupported declaration testimony as a “naked assertion” and distinguished declarations supported by explanatory detail or evidence. 859 F.3d 1341, 1345–48 (Fed. Cir. 2017). A SMED that merely states that claims improve technology is vulnerable for the same reason.

The examiner memorandum itself points in the same direction: objective evidence should be supported by actual proof, and legal conclusions are not entitled to weight even if the underlying factual basis may be persuasive. Examiner SMED Memo 3. A strong SMED should therefore be corroborated by contemporaneous technical documents, RFCs, standards, architecture diagrams, system logs, source-code excerpts, benchmarking data, design records, technical publications, screenshots of operative workflows, or independent expert testimony.

Corroboration serves at least three functions. It reduces the risk that the declaration will be dismissed as interested advocacy. It converts broad technical characterizations into specific facts. And it makes the prosecution record more useful if the patent later faces a Rule 12 motion, summary judgment motion, post-grant proceeding, or appeal. The declaration should not ask the examiner or later court to take the declarant’s word for the system’s technical character. It should guide the decisionmaker through objective evidence showing how the claimed architecture works.

V.          Case Study One: Email-Based E-Commerce and the Authentication Control Packet

The first declaration, filed in Application No. 17/883,539 for “Email-Based E-Commerce,” provides a useful example of how a SMED can reframe an examiner’s high-level characterization. The eligibility risk is obvious: an examiner can characterize the claims as advertising, marketing, or selling goods by email. Courts may be receptive to that high-level framing if the claims appear to recite little more than a commercial transaction carried out with generic computing components.

The declaration responds by identifying a lower-level technical problem: how an e-commerce system can determine whether a returned SMTP order email generated from a mailto hyperlink should be trusted for order execution when the message arrives outside a continuous, server-controlled checkout session. Declaration of John P. Killoran, Jr. Under 37 C.F.R. § 1.132 ¶¶ 199–201, U.S. Patent Application No. 17/883,539 (May 9, 2026) (“’539 SMED”). 

The declaration’s theory is not that SMTP, mailto links, UUIDs, processors, databases, or network interfaces were newly invented. Rather, it asserts that the claimed improvement lies in a particular ordered architecture: storing server-side authentication context, including a UUID in a mailto-generated transaction path, receiving a returned SMTP order email, parsing returned fields, validating those fields against a stored authentication control packet, enforcing expiration, and conditionally gating order execution. ’539 SMED ¶¶ 153–58.

That framing matters under BASCOM. The declaration does not rely on novelty of each component in isolation. Instead, it emphasizes the ordered arrangement of known components and the resulting change in system operation. BASCOM, 827 F.3d at 1349–52. The same strategy is consistent with Enfish and SRI because it attempts to identify a specific improvement in how the computer or networked transaction system operates, rather than merely implementing a commercial practice on generic technology. Enfish, 822 F.3d at 1335–39; SRI, 930 F.3d at 1303–04.

This is also where the strategy is vulnerable. If the claim language does not actually require the authentication-control architecture, the declaration cannot supply it. If the claim merely recites receiving an email and processing an order, the examiner or a court may discount the declaration as describing an embodiment rather than the claimed invention. The strength of the SMED therefore depends on the correspondence between the asserted technical architecture and the actual claim limitations.

The declaration is strongest where it uses corroboration. It cites SMTP and mailto standards to explain that SMTP transports email and that a mailto URI composes a message, but neither supplies the claimed server-side authentication control packet, field-by-field verification, or expiration gate. ’539 SMED ¶¶ 90–94. It also cites RFC 4122 for the principle that UUID possession alone should not be treated as a security capability, and uses OWASP guidance to explain the technical difference between continuous web-session context and the claimed SMTP/mailto transaction path. ’539 SMED ¶¶ 195–98.

The technical story is simple but not automatic: the returned email is not self-authenticating. The system does not proceed directly from receipt of email to execution of the order. The returned message is parsed and authenticated against an expiration-bound, server-side authentication control packet before order execution. ’539 SMED ¶¶ 198–203.

For future SMED practice, this case study teaches that the declaration should identify the trust boundary. Here, the trust boundary sits between receipt of the returned SMTP order email and execution of the transaction. The claim-tethered technical improvement is the mechanism that prevents an email from being treated as a valid order merely because it was received.

VI.       Case Study Two: Vendor Token Generator and Stateful Token Security

The second declaration, filed in Application No. 19/024,922 for “Vendor Token Generator,” is a cleaner model of modern SMED practice. It expressly states that the declaration is submitted only for § 101 and not for novelty, obviousness, written description, enablement, definiteness, or any other patentability issue. Declaration of John P. Killoran, Jr. Under 37 C.F.R. § 1.132 ¶¶ 1–3, U.S. Patent Application No. 19/024,922 (May 30, 2026) (“’922 SMED”). That separation tracks the USPTO’s recommendation that applicants file separate SMEDs to avoid intermingling eligibility evidence with § 102, § 103, or § 112 evidence. Apr. 2026 SMED Memo 3–4.

The declaration’s core theory is that the claims are not directed merely to using email to conduct a purchase. Instead, they recite an ordered security architecture for SMTP-based transaction messages containing tokenized transaction instructions. The architecture includes receiving an SMTP message containing a token, decoding the token, validating token freshness, checking one-time-use status, recording token use, authenticating the sender against decoded token information, verifying that the vendor is authorized to generate the token, and performing the transaction only when those security checks succeed. ’922 SMED ¶¶ 201–05.

This declaration is especially strong because the technical improvement is corroborated by multiple objective concepts familiar in security engineering: expiration, nonce or one-time-use validation, token-use datastore state, cryptographic or MAC-based vendor authentication, key validity, sender-token matching, and conditional execution. The declaration explains that token decoding allows the e-commerce system to recover customer, vendor, and transaction information from the token rather than relying only on visible email text or apparent sender information; expiration validation addresses stale-token risk; nonce or one-time-use validation addresses replay risk; token-use recording creates persistent state; sender authentication binds the SMTP message to decoded customer identity; and vendor-authenticator verification binds the token to an authorized vendor. ’922 SMED ¶¶ 203–05.

The declaration also includes a nexus table mapping claim limitations to technical functions and representative specification support. That table identifies the mailto-generated message, token generated for the mailto link, vendor identifier, transaction data, expiration time, nonce or one-time-use value, token-use datastore, sender authentication, vendor-authenticator verification, MAC implementation, and checksum-based replay prevention. ’922 SMED ¶¶ 195–200.

That kind of nexus table is important. It makes the examiner’s job easier, but it also makes the record more litigation-ready. If a later court asks whether the alleged improvement is actually claimed, the table gives the applicant a concise record showing the correspondence between claim language, technical function, and specification support. That is precisely the discipline required by ChargePoint, Simio, and American Axle: the improvement must be in the claims, not merely in the specification or in an expert’s characterization.

This declaration also better anticipates the evidentiary-weight problem. It says the declarant is not offering a legal conclusion on eligibility, but is instead providing technical testimony about how a POSITA would understand the specification and claims. ’922 SMED ¶¶ 201–03. That is consistent with the examiner memorandum’s instruction that legal conclusions are not entitled to weight, while underlying factual bases may be persuasive. Examiner SMED Memo 3.

Still, even this stronger declaration is not immune from attack. An examiner or court may say that expiration checks, nonce validation, token recording, sender matching, and cryptographic verification are familiar security concepts applied to an e-commerce setting. The answer cannot simply be that those concepts are technical. The stronger answer is that the claim recites a particular ordered combination adapted to SMTP-based tokenized transaction messages, and that the specification supports that architecture as a change in how the e-commerce system processes such messages before execution.

Here too, corroboration can be further strengthened. The declaration already points to technical standards and security concepts. Future iterations could add vendor-key lifecycle records, token-generation diagrams, sample encoded and decoded token fields, datastore schema excerpts, pseudocode for nonce recording, replay-rejection logs, MAC verification examples, or independent security expert testimony. Those materials would help avoid the EmeraChem problem: the declaration would not merely assert that the system is secure or technological; it would demonstrate the specific system behavior through objective evidence. EmeraChem, 859 F.3d at 1345–48.

VII.   Conclusion

SMEDs are valuable because they allow applicants to build a factual record for § 101. They can explain what a POSITA would have understood from the specification and claims, identify the technological problem, describe the technical mechanism, rebut unsupported conventionality or mental-process assertions, and show why the ordered claim combination changes system operation.

But SMEDs are bounded tools. The claims must recite the technological improvement. The specification must support it as filed. The declaration must have a nexus to the claims. And the testimony should be corroborated.

The two Swoop declarations illustrate the right direction. The Email-Based E-Commerce declaration identifies a UUID-linked, expiration-bound, server-side authentication control packet that changes how a returned SMTP/mailto order email is treated before order execution. The Vendor Token Generator declaration identifies a stateful, cryptographically verifiable, sender-bound token-security architecture for SMTP-based transaction messages. Both declarations move the analysis away from the high-level commercial result and toward the claimed system behavior.

That is the core of good SMED practice: do not ask the declarant to rescue eligibility. Use the declarant to prove, with claim-tethered and corroborated technical facts, that the claimed invention was technological from the beginning.