The Alice/Mayo two-step test is the Supreme Court framework that patent examiners and federal courts use to decide whether a claim is eligible for patent protection under 35 U.S.C. § 101. It works by asking two questions in order. First, is the claim directed to a law of nature, a natural phenomenon, or an abstract idea? Second, if it is, does the claim add an “inventive concept” that amounts to significantly more than the excluded concept itself? A claim that clears either question is eligible. A claim that fails both is not.
The test is the single biggest eligibility hurdle in patent prosecution today, and it hits software and diagnostic claims hardest.
Where the Test Comes From
Two Supreme Court decisions built the framework.
In Mayo Collaborative Services v. Prometheus Laboratories (2012), the Court struck down patents on a diagnostic method that told doctors to measure metabolite levels in patients taking thiopurine drugs and adjust dosages based on the results. The claims did nothing more than describe a natural correlation between metabolite levels and drug effect, then instruct practitioners to “apply it” with routine lab techniques. Those extra steps were “well-understood, routine, conventional activity previously engaged in by researchers in the field” and did not turn the natural law into a patentable application.1Justia Law. Mayo Collaborative Services v. Prometheus Laboratories, Inc.
Two years later, Alice Corp. v. CLS Bank International (2014) extended the same framework to abstract ideas. The patents there claimed a method of using a third-party intermediary to reduce settlement risk in financial transactions, plus computer systems to run it. The Court found the claims directed to the abstract idea of intermediated settlement, “a fundamental economic practice long prevalent in our system of commerce.” Running it on a generic computer did not save the claims: “Stating an abstract idea, adding the words ‘apply it with a computer,’ simply combines two steps, with the same deficient result.”2Justia Law. Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014)
The two decisions locked in the framework that examiners and courts now apply to any claim that might touch a judicial exception.
Step One: Is the Claim Directed to a Judicial Exception?
Step One asks whether the claim is directed to a law of nature, a natural phenomenon, or an abstract idea. These three categories are called judicial exceptions because the Supreme Court created them, not Congress. Section 101 itself allows patents on “any new and useful process, machine, manufacture, or composition of matter.”3Office of the Law Revision Counsel. 35 USC 101 – Inventions Patentable The Court has held that the three exceptions sit outside those boundaries no matter how novel or useful a particular application might be.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
The USPTO splits Step One into two prongs.
Prong One asks whether the claim recites a judicial exception at all. Abstract ideas fall into three groupings: mathematical concepts, methods of organizing human activity (commercial interactions, managing personal relationships, following rules), and mental processes that a person could perform in their head or with pen and paper. Laws of nature include mathematical relationships describing physical phenomena. Natural phenomena cover things like naturally occurring organisms, minerals, and genetic sequences.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
If the claim recites no judicial exception, the analysis ends and the claim is eligible. A new mechanical tool that works through physical structure, for example, clears this step without further scrutiny.
The Mental Process Trap
The mental process category catches more claims than applicants expect. The USPTO treats any concept that could be performed through observation, evaluation, judgment, or opinion as a potential mental process. Running that operation on a computer does not change the analysis. If a person could do it in their head and the computer is just executing it faster, the claim still recites an abstract idea. USPTO examples include collecting information, analyzing it, and displaying results; comparing genetic sequences to identify alterations; and identifying head shapes and applying hair designs.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
The line matters. Operations a human mind genuinely cannot perform are not mental processes. Detecting suspicious network activity by analyzing live packet data, calculating GPS positions from satellite signals, and performing multi-step data encryption have all been found to fall outside the category because no person could practically do them without a machine.
The Practical Application Escape
If a claim does recite a judicial exception, Prong Two asks whether the claim as a whole integrates that exception into a “practical application.” Many claims survive here. If the additional elements use the exception in a meaningful way, the claim is not considered “directed to” it and passes without ever reaching Step Two.
The USPTO recognizes several ways a claim can integrate a judicial exception into a practical application:
- Improving the functioning of a computer or another technical field.
- Applying the exception to a particular medical treatment or prevention of a disease.
- Using a specific machine that is integral to the invention, not just a generic computer.
- Transforming a particular article into a different state or thing.
- Applying the exception in some other meaningful way that goes beyond linking it to a technological environment.
Certain additions do not count. Saying “apply it” or “use a computer,” bolting on data-gathering steps only tangentially related to the core invention, or loosely tying the exception to a particular field of use will not carry a claim through this prong.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
The Federal Circuit’s decision in Enfish, LLC v. Microsoft Corp. (2016) illustrates how far Prong Two can go. The court held that software claims directed to a self-referential database table were not abstract because they improved how computers actually operate. “Software can make non-abstract improvements to computer technology just as hardware improvements can.” When the focus of a claim is a specific improvement to computer capabilities rather than using the computer as a tool to carry out an abstract process, the claim can survive at Step One.5Justia Law. Enfish, LLC v. Microsoft Corp., No. 15-1244 (Fed. Cir. 2016)
Step Two: The Search for an Inventive Concept
Claims that fail both prongs of Step One reach Step Two, where the examiner looks for an “inventive concept.” The question is whether the claim adds “significantly more” than the judicial exception itself. As the Supreme Court put it in Mayo: “to transform an unpatentable law of nature into a patent-eligible application of such a law, a patent must do more than simply state the law of nature while adding the words ‘apply it.’ It must limit its reach to a particular, inventive application of the law.”1Justia Law. Mayo Collaborative Services v. Prometheus Laboratories, Inc.
The examiner looks at the claim elements individually and as a combined sequence. A claim can contain several conventional steps that together produce a non-conventional arrangement, and the combination can supply the inventive concept even when no single element could. If every element is standard and the combination adds nothing new, the claim fails. The concern driving this analysis is preemption: patenting a bare abstract idea or law of nature would block everyone else from using a fundamental building block.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
Pass Step Two and the claim is eligible. Fail it and the examiner rejects the claim under 35 U.S.C. § 101.
What Fails the Inventive Concept Requirement
Courts and examiners have built a clear picture of what does not qualify as “significantly more.”
The most common failure is implementing an abstract idea on a generic computer. In Alice, the Court found the computer components performed “purely conventional” functions: creating and maintaining accounts, obtaining data, adjusting balances, and issuing automated instructions. None of them purported to improve the functioning of the computer itself.2Justia Law. Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014)
The USPTO has identified computer functions that courts have consistently treated as conventional when claimed generically:
- Transmitting or receiving data over a network.
- Performing repetitive calculations.
- Electronic recordkeeping.
- Storing and retrieving data in memory.
- Scanning or extracting data from documents.
Standard laboratory techniques face similar treatment in life sciences. Measuring biomarker levels in blood by standard means, using PCR to amplify DNA, and detecting enzymes in a sample have all been held insufficient to supply an inventive concept when they merely carry out a law of nature.4United States Patent and Trademark Office. MPEP 2106 – Patent Subject Matter Eligibility
The point is not that these techniques are unimportant. They are foundational, and that is exactly the problem. Claiming a newly discovered natural correlation plus “measure it the standard way” would give one patent holder a lock on all practical uses of the discovery. The inventive concept requirement forces applicants to show something genuinely new in how the discovery is applied.
Conventionality Is a Factual Question
Whether a claim element is “well-understood, routine, and conventional” is not a pure legal call. Two Federal Circuit decisions in 2018 established that it is a factual question, which changes how eligibility fights actually play out.
In Berkheimer v. HP Inc., the court held that the challenger bears the burden of showing conventionality and “there must be evidence supporting a finding that the additional elements were well-understood, routine, and conventional.”6Justia Law. Berkheimer v. HP Inc. The companion case, Aatrix Software, Inc. v. Green Shades Software, Inc., fleshed out the kind of evidence that matters. The patent specification is a key source: if it describes the technology as an improvement or as unconventional, that cuts in favor of eligibility. If the specification admits the technology is routine, the patent holder will struggle to argue otherwise later. Expert testimony can help, but vague declarations that features were “known functions” carry little weight. Courts want specifics.7Justia Law. Aatrix Software, Inc. v. Green Shades Software, Inc.
The practical consequence: before Berkheimer and Aatrix, courts sometimes dismissed eligibility challenges on the pleadings by treating conventionality as pure law. Now, a patent holder who can point to a genuine factual dispute about whether the claimed technology was routine at the time of filing can survive a motion to dismiss. Drafters who describe the technical problem and how the invention solves it in a non-obvious way create a record that pays off later.
How the Test Hits Different Industries
Software and Computer-Implemented Inventions
Software claims bear the heaviest burden. Because software processes data according to algorithms, nearly any software claim can be characterized as implementing a mathematical concept or mental process. Financial transactions, data analysis, user interface workflows, and e-commerce methods routinely face Section 101 rejections. Alice itself involved software, and its language about generic computer implementation has been applied broadly across the technology sector.
Enfish offers the clearest path forward: frame claims around how the invention improves computer functionality rather than what abstract task it accomplishes. A claim to a new database structure that makes computers faster is fundamentally different from a claim to a business method that happens to run on a database.
Life Sciences and Diagnostics
Diagnostic methods and biotechnology face scrutiny under the law-of-nature and natural-phenomena exceptions. In Association for Molecular Pathology v. Myriad Genetics (2013), the Supreme Court held that naturally occurring DNA segments are not patentable merely because they have been isolated, though synthetically created cDNA remains eligible because it does not occur in nature.8U.S. Reports. Association for Molecular Pathology v. Myriad Genetics, Inc.
Diagnostic methods that identify a correlation between a biomarker and a disease face an especially difficult path. After Mayo, a claim that says “measure this metabolite level and conclude the patient has condition X” is directed to a law of nature, and adding standard testing steps does not supply an inventive concept. Companies developing personalized medicine and companion diagnostics have to build claims around specific, unconventional testing methods or novel technological implementations of the diagnostic process rather than the correlation itself.
How the Test Applies to AI Inventions
In July 2024, the USPTO issued updated guidance addressing how the Alice/Mayo test applies to artificial intelligence inventions. The guidance does not change the two-step framework; it directs examiners on how to apply it to AI claims.9Federal Register. 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence
Two points matter most. First, the fact that an invention was created with the assistance of AI has no bearing on the eligibility analysis. The test looks at the claimed invention, not how the inventor developed it. Whether a human conceived the idea independently or used an AI tool along the way, the Section 101 analysis is identical. The separate question of whether an AI system can be named as an inventor is a different issue; current law requires at least one natural person to have made a significant contribution.
Second, the guidance added examination examples for AI claims, including a neural network for anomaly detection, an AI method for separating speech from background noise, and an AI model for personalizing medical treatment. They illustrate the same distinction that runs through the whole test: an AI claim that recites a mathematical concept at a high level of generality is abstract, but one that shows a specific improvement to computer functionality or solves a concrete technical problem can pass the practical application analysis at Step One.
The Test May Not Last Forever
The Patent Eligibility Restoration Act (PERA), reintroduced in the Senate in May 2025, would replace the Alice/Mayo framework entirely. As of mid-2025, the bill has been introduced but not enacted.10Congress.gov. S.1546 – Patent Eligibility Restoration Act of 2025
PERA would keep the existing statutory categories (process, machine, manufacture, composition of matter) but limit exclusions to four defined categories rather than the open-ended judicial exceptions:
- Standalone mathematical formulas not part of a claimed invention.
- Processes that are substantially economic, financial, business, social, cultural, or artistic, even if a claim step mentions a machine.
- Processes performed solely in the human mind, or occurring naturally without human activity.
- Unmodified human genes as they exist in the body, and unmodified natural materials as they exist in nature.
PERA would also target the drafting technique of adding token computer steps to save an otherwise ineligible claim, specifying that pre- or post-solution computer activity is not enough if the computer is not actually necessary to perform the invention. If enacted, PERA would eliminate the two-step analysis and replace it with these defined boundaries. Previous versions did not advance, so whether the 2025 bill fares differently remains open.
Until then, the Alice/Mayo two-step continues to govern every Section 101 analysis in the United States.