How to respond to a 35 U.S.C. 101 rejection

What the examiner must show

35 U.S.C. 101 covers any new and useful process, machine, manufacture or composition of matter. The courts read in three exceptions: abstract ideas, laws of nature and natural phenomena. The Supreme Court's two-step test from Mayo v. Prometheus (2012) and Alice v. CLS Bank (2014) decides whether a claim falls into one of them, and the USPTO's version of that test, first issued as the 2019 Revised Patent Subject Matter Eligibility Guidance, now lives in MPEP 2103 to 2106.07.

The burden starts with the examiner. Under MPEP 2106.07(a) the rejection must identify the exception, explain why it falls into a recognized grouping, identify the additional elements and explain why they do not integrate the exception or add significantly more. Every patentability rejection has to be established by a preponderance of the evidence. The August 4, 2025 memo adds that in a close call the examiner should reject only when it is more likely than not (more than 50%) that the claim is ineligible, and not simply because the examiner is unsure. Once a proper rejection is made, the burden shifts to you to amend, argue or submit evidence (MPEP 2106.07(b)).

Prong one: does the claim recite an exception?

The examiner may only call something an abstract idea if it fits one of the three groupings in MPEP 2106.04(a)(2): mathematical concepts (relationships, formulas, calculations), certain methods of organizing human activity (fundamental economic practices, commercial interactions, managing personal behavior), and mental processes (observation, evaluation, judgment, opinion). Two limits on prong one are worth knowing by heart.

First, a claim must recite the exception, not merely involve it. The 2025 memo contrasts USPTO Example 39, where "training the neural network in a first stage using the first training set" does not recite math, with claim 2 of Example 47, which names backpropagation and gradient descent and therefore does. Second, the mental process grouping covers only steps that can practically be performed in the human mind or with pen and paper. The memo tells examiners not to stretch it to limitations the human mind is not equipped to perform, and says claim limitations that encompass AI in that way fall outside the grouping.

Prong two: is the exception integrated into a practical application?

MPEP 2106.04(d) lists what can integrate an exception: an improvement in the functioning of a computer or another technology; a particular treatment or prophylaxis for a disease; use with a particular machine integral to the claim; a transformation of an article to a different state; or another meaningful limitation beyond linking the exception to a field. It also lists what does not: "apply it" or mere instructions to run the idea on a computer (2106.05(f)), insignificant extra-solution activity such as data gathering or outputting a result (2106.05(g)), and limiting the idea to a technological environment or field of use (2106.05(h)). The analysis looks at the claim as a whole, including how the additional elements interact with the exception, not at each element in isolation.

Step 2B and Berkheimer

If the claim is still "directed to" the exception, Step 2B asks whether the additional elements add an inventive concept. Most of the considerations repeat prong two. The one that is new is whether an element is well-understood, routine and conventional (MPEP 2106.05(d)). After Berkheimer v. HP (Fed. Cir. 2018) that is a factual finding, and MPEP 2106.07(a), subsection III requires the examiner to support it with one of four things: an express statement in your specification or prosecution history, a court decision noting the element is conventional, a publication, or official notice. Silence in the specification about an element's details is not enough on its own. Remember too that conventional parts can still add up: under BASCOM v. AT&T Mobility (Fed. Cir. 2016), cited in MPEP 2106.05, an inventive concept may lie in a non-conventional arrangement of individually conventional components.

What changed from 2024 to 2026

The current MPEP is still the Ninth Edition, Revision 01.2024, published November 2024. Everything since then has come out as guidance, memos and an advance notice of MPEP changes, all listed on the USPTO subject matter eligibility page (last updated September 29, 2026). Cite them by name and date in your remarks.

Ex parte Desjardins, in practical terms

The claim in Desjardins trained a machine learning model on a second task while "protecting performance of the machine learning model on the first machine learning task", a fix for what the specification called catastrophic forgetting. The Appeals Review Panel agreed the claim recited a mathematical concept at prong one. It then held at prong two that the specification identified improvements to how the model itself operates (reduced storage, reduced system complexity, preserved performance on earlier tasks) and that the claim, read as a whole, reflected them. The claim was therefore not directed to the exception.

The December 5, 2025 advance notice writes three points into MPEP 2106.04(d)(1) and 2106.05(a) that you can quote back to an examiner. Claims should not be evaluated "at such a high level of generality" that meaningful technical limitations are dismissed without explanation. Additional elements should not be written off as "generic computer components" without asking whether they confer a technological improvement. And the claim does not have to state the improvement in words; it has to include the components or steps that provide it.

How to read the rejection

Before deciding anything, mark up the action against the framework. A rejection that skips a step, or takes one on assertion alone, is the easiest kind to traverse.

1. Which words are the exception? The examiner should quote or identify the limitations said to recite it and name the grouping. "The claim is directed to collecting and analyzing data" without a grouping is a gap.
2. Is the grouping right? For a mental process, ask whether the step can practically be done in the head. Processing thousands of sensor readings per second, or adjusting model weights, usually cannot.
3. What was labeled "additional"? Check whether limitations that carry the invention were folded into the abstract idea so they never get weighed at prong two.
4. Was the specification read? MPEP 2106.04(d)(1) requires the examiner to consult the specification for a described improvement. If the action never mentions it, say so.
5. Is Step 2B supported? Look for one of the four forms of Berkheimer support for every "conventional" finding.
6. Were dependent claims analyzed? A blanket statement that the dependent claims "do not add significantly more" often hides one that does.

Argue, amend, or both

Argue alone when the claim already contains the technical steps and the examiner has generalized them away, misplaced a step in a grouping, or made an unsupported Step 2B finding. Amend when the claim is broad enough to read on the idea of the result: "determining", "analyzing" and "outputting" with no account of how. Most strong responses do both: a narrowing amendment that pulls in the technical mechanism from the specification, and an argument tying that mechanism to the improvement.

Every amendment needs written support in the application as filed, and the same amendment usually has to answer the art rejections too. Choose limitations that do double duty: a concrete technical step the cited references lack is the best amendment for 101, 102 and 103 at once. See the guides to 102 rejections, 103 rejections and the 112 support requirements that limit what you can add.

These snippets are illustrations of structure, not language to file. The paragraph numbers and facts are invented.

Building the technological improvement argument

The improvement consideration is a two-part test, and both parts come from the record you already have. Part one is the specification: it must describe the invention so that a skilled person would recognize an improvement in computer function or another technology. A bare assertion such as "the system is faster and more efficient" does not count (MPEP 2106.04(d)(1)). Part two is the claim: it must include the components or steps that provide the improvement, which is where Intellectual Ventures v. Symantec claims failed, because the claims lacked the limitations that addressed the problems the patent owner cited.

The cases the MPEP lists as improvements are useful anchors: Enfish (a self-referential database table), McRO (rules for automated lip-sync animation), Visual Memory (a configurable memory system), Finjan (behavior-based virus scanning), SRI v. Cisco (network monitors analyzing packets), and now Desjardins. The common thread is a particular way of achieving a result, as opposed to claiming the result. The 2025 memo gives examiners three questions to ask, and your argument should answer each: does the claim recite how the solution is accomplished; is the computer a tool for an existing process or is computer capability itself improved; and how particular is the application of the exception?

For a fintech or business-method claim, the argument is harder because the improvement is often to the business, not the technology. Look for the technical constraint the invention works around (bandwidth, latency, concurrency, storage, security of a data channel) and argue that. If the specification only describes business benefits, no argument will repair it, and a continuation with narrower claims or an appeal on a cleaner record may be the realistic route.

Declarations under Rule 132 (SMEDs)

Evidence offered to traverse a rejection goes in by declaration under 37 CFR 1.132, and MPEP 2106.07(b) expressly allows it for eligibility. Since December 2025 the USPTO has called these Subject Matter Eligibility Declarations (SMEDs), and its April 30, 2026 and September 29, 2026 memos set out what it wants to see. A SMED is optional. It is useful when the dispute is about a fact rather than the law.

• Good subjects: why a step cannot practically be performed in the human mind; what the state of the art was at filing and why a skilled reader would see the disclosed technique as a technical improvement; whether an element was in fact conventional.
• File it separately from declarations on 103 or 112, using the Patent Center document code AF/D.SMED and the description "Subject Matter Eligibility Declaration", so the issues do not get tangled.
• File it early. It can be filed any time before prosecution closes, even before the first action. After a final rejection it is admitted only in limited circumstances (MPEP 716.01).
• Prefer a disinterested declarant and give factual support, such as a trade article from before the filing date describing the problem. An inventor's declaration counts, but carries less weight (MPEP 716.01(c)).
• Say which step it answers, for example "submitted to show that limitation (c) cannot practically be performed in the human mind (Step 2A, prong one)". An interview to walk the examiner through it helps.

A declaration cannot add an improvement the specification never described. It can only help the examiner see what a skilled person would have understood from it.

Common mistakes

• Arguing that the claim is novel or non-obvious. MPEP 2106.05 says the search for an inventive concept should not be confused with a novelty or non-obviousness determination. Keep 102 and 103 arguments in their own sections.
• Reciting the improvement only in the remarks. Under MPEP 2106.04(d)(1) the specification has to describe it in enough detail that a skilled person would see it, and the claim has to contain the steps that deliver it.
• Adding hardware words without changing what the claim does. "Using a processor" or "on a server" is the textbook "apply it" limitation of MPEP 2106.05(f).
• Letting a Step 2B finding of "well-understood, routine, conventional" go unchallenged when the examiner cited nothing for it. That finding needs one of the four forms of support in MPEP 2106.07(a), subsection III.
• Ignoring the dependent claims. Each one gets its own analysis, and a dependent claim with a concrete technical step is often the fastest route to an allowable independent claim.
• Filing a declaration that only states the conclusion. A SMED is weighed for the facts behind the opinion, the interest of the declarant and any opposing evidence (MPEP 716.01(c)).

The best 101 response is written before filing: a specification that names the technical problem, explains the mechanism in detail, and claims that recite the mechanism rather than the outcome. See how to write patent claims for structure and fallback positions.

More USPTO guides

• How to file a provisional patent application (US)

  What a US provisional must contain, the USPTO fees, the 12-month deadline under 35 U.S.C. 119(e), and when to skip it.

• USPTO patent fees (2026)

  Every USPTO patent fee for large, small and micro entities, from filing to maintenance, with a fee calculator.

• Patent drawing requirements: USPTO, EPO and India

  37 CFR 1.84, the EPO rules that replaced Rule 46 EPC, and Indian rule 15, compared side by side.

• How to write patent claims: USPTO, EPO and India

  Claim structure and worked examples, then the claim rules and fees at the USPTO, the EPO and the Indian Patent Office, side by side.

• How to respond to a USPTO office action

  Reply deadlines and extension fees, 101, 102, 103 and 112 rejections, interviews, and options after a final rejection.

• Types of patent searches, and which one you need

  Patentability, freedom-to-operate, invalidity and landscape searches compared by purpose, timing, scope and deliverable.

• Patentability search: how to check novelty before you file

  What counts as prior art under 35 U.S.C. 102 and Art. 54 EPC, the grace period gap, search steps and how to read results.

• Freedom-to-operate search in the US and Europe

  In-force claims only, legal status checks at the USPTO, EPO and UPC, FTO opinions after Halo, and design-arounds.

• Invalidity search: prior art to challenge a patent

  IPR, PGR, EPO opposition and UPC revocation deadlines, prior art dates, standards documents and claim charts.

• Patent landscape analysis, step by step

  Building a CPC and keyword dataset, cleaning families and names, the metrics that matter and the 18-month publication lag.

• How to use USPTO Patent Public Search

  Basic and Advanced Search, the three databases, field codes, ADJ, NEAR, WITH and SAME, truncation and a worked search.

• Google Patents search guide

  Search operators and fields, Boolean and proximity syntax, non-patent literature, and where Google Patents falls short of the official databases.

• The Cooperative Patent Classification (CPC), explained

  How CPC symbols are built, CPC against IPC, the Y section, how to find the right class and how to search it in each database.

• Patent search Boolean operators and search strings

  Boolean, proximity and truncation syntax compared across Patent Public Search, Espacenet, Google Patents and The Lens, with semantic search for contrast.

• How to respond to a 35 U.S.C. 102 rejection

  Anticipation under MPEP 2131, AIA 102(a)(1) and (a)(2), the 102(b) exceptions, inherency and Rule 130 declarations.

• How to respond to a 35 U.S.C. 103 rejection

  Graham and the KSR rationales in MPEP 2143, teaching away, motivation to combine, secondary considerations with nexus, and Rule 132 evidence.

• How to respond to a 35 U.S.C. 112 rejection

  Written description and enablement after Amgen, 112(b) indefiniteness, 112(f) means-plus-function after Williamson, and 112(d) dependent claims.

• The PCT application: timeline, fees and national phase

  The international phase month by month, receiving Office and ISA fees in CHF, USD and EUR, national phase deadlines by office, and a deadline calculator.

• Entering the US national stage under 35 U.S.C. 371

  The 30-month deadline and revival, the 371 requirements, national stage fees by entity size, and 371 versus a bypass continuation.

• The US foreign filing license (35 U.S.C. 184)

  When a US-made invention needs a license to be filed abroad, the six-month rule, expedited and retroactive licenses, and the Indian and European equivalents.

• Filing a US patent from Europe or India

  The 2026 US practitioner rule for foreign applicants, Paris priority and certified copies, small entity status, PCT or direct, fees and translations.

• What is a utility patent?

  What a US utility patent covers under 35 U.S.C. 101, the parts of the application in 37 CFR 1.77 order, the 20-year term and USPTO costs.

• Design patents in the United States

  What a design patent protects, drawings and the single claim, the 15-year term, Hague filings, LKQ v. GM, and design vs utility patents.

• How to file a nonprovisional patent application

  Required parts and fees, claiming a provisional within 12 months, DOCX filing in Patent Center, and what happens after you file.

• Provisional vs nonprovisional patent application

  The two US filings side by side, the cost of each route, the effect on term and publication, and a guide to which to file.

• Continuation, CIP and divisional applications

  Copendency, patent term, double patenting and terminal disclaimers, the 2025 continuing application fee, and continuation strategy.

• How much does a patent cost?

  USPTO fees by entity size from filing to the last maintenance fee, what drives professional fees, EPO fees, and an estimator.

• How to patent an idea in the US

  Nine steps from patentability and search to provisional filing, Patent Center, examination and maintenance, with the USPTO's free help programs.

• Small entity vs micro entity status

  Who qualifies under 37 CFR 1.27 and 1.29, the 2026 income limit, SB/15A and SB/15B, license traps, fixing errors, and an eligibility checker.

• How long does a patent last?

  US utility and design terms, patent term adjustment and extension, terminal disclaimers, maintenance fee deadlines, Europe and India, with an expiry estimator.

• Can AI write a patent application?

  What AI drafting does well and badly, the USPTO and EPO positions on AI tools, AI inventorship after 2025, and confidentiality.

• Restriction requirements and election of species

  How to answer a USPTO restriction or election of species requirement, from traverse and rejoinder to divisionals and unity in 371 cases.

• Options after a final rejection

  Amendment after final, advisory actions, the end of AFCP 2.0, RCE, pre-appeal and appeal, with fees, timing and USPTO outcome data.

• Request for continued examination (RCE)

  What 37 CFR 1.114 requires, first and later RCE fees, patent term effects, and when an RCE beats an appeal.

• Information disclosure statements (IDS)

  The duty of disclosure, IDS timing windows and fees, the 2025 size fee, EPO search reports, Therasense and QPIDS.

• PTAB ex parte appeals

  Notice of appeal to decision step by step, with fees, deadlines and the PTAB's FY 2026 pendency and outcome statistics.

• Track One prioritized examination

  Track One requirements, fees by entity size, the 20,000 annual cap and USPTO timing data, plus PPH and age or health petitions.

• 35 U.S.C., annotated: US patent law section by section

  The current text of 34 key sections of Title 35 with plain-English summaries, key cases, 37 CFR rules and MPEP references.

All guides, by patent office