Many companies have valuable inventions buried in their codebase, manufacturing process, or hardware design without realizing it. They often emerge through routine product development: an engineer solves a performance problem, a team develops a more reliable testing method, or a product team designs an unusual configuration that gives the product a competitive edge.
But all these solutions can be commercially important. They may protect the technical advantage that makes the product faster, more reliable, or harder for a competitor to replicate. Identifying them early — before a product launch, conference presentation, or publication — is one of the more valuable things a technical team can do.
This guide covers the legal requirements for patentability in the United States, examples of what does and does not qualify, and a practical framework for finding the inventions your team has already built.
What are the requirements for a patentable invention?
Under U.S. patent law (35 U.S.C. §§ 101–103), an invention must meet four requirements to be patentable: eligible subject matter, novelty, non-obviousness, and utility.
Eligible subject matter (Section 101)
The invention must fall into one of the categories the patent statute covers: a process, machine, manufacture, or composition of matter. Most tangible inventions clear this bar without difficulty; however, some categories of invention — particularly software, business methods, and biotechnology — require more careful analysis.
This means that a new chemical compound is eligible subject matter. A new algorithm that reduces server response time by restructuring how queries reach a database may be eligible, but the application must describe the specific technical improvement — not just the result it produces.
Novelty (Section 102)
The invention must be new. If someone else has already described, patented, or publicly used the same invention before your priority date, it isn't novel. This is where prior art comes in — the body of existing knowledge against which your invention is measured.
Novelty doesn't require that your invention be completely unrelated to everything that came before it. It means no single prior art reference describes every element of your claimed invention. A new combination of existing components may still be novel if no prior reference disclosed that particular combination.
One common trap is that your own public disclosures count as prior art. If your CTO presented the core technical approach at a conference, published it in a blog post, or disclosed it through an open-source repository, that disclosure starts the clock. The U.S. offers a one-year grace period for an inventor's own disclosures, but most other countries do not.
Non-obviousness (Section 103)
Even if your invention is novel, it still has to be non-obvious. This is the requirement that often trips up applications. The question isn't whether someone has built exactly what you built, but whether a person with ordinary skill in your technical field would have found it obvious to combine or modify existing knowledge to reach the same result.
In practice, this means that small, incremental improvements to well-known approaches are hard to patent. Merely adding a standard caching layer to a standard web application probably won't clear the bar. Developing a novel approach to cache invalidation that solves a previously unsolved consistency problem may well clear it.
Generally, the stronger your evidence that skilled practitioners wouldn't have arrived at the same solution through routine experimentation, the better your non-obviousness argument. Prior failed attempts, unexpected results, and adoption by competitors are all factors that can support a non-obviousness case.
Utility (Section 101)
The invention must be useful. This is the easiest requirement to meet; almost anything with a practical application satisfies it. The utility requirement mostly filters out purely theoretical concepts without any real-world application — not a concern for the vast majority of companies building products.
What types of inventions can be patented?
The patent statute defines four broad categories of patentable subject matter. Most inventions fit into at least one, and many fit into more than one.
Processes and methods
This includes a specific sequence of steps that produces a useful result. Manufacturing processes, chemical synthesis methods, data processing pipelines, testing protocols, and treatment methods can all qualify. Examples may include:
- A multi-step method for allocating computing resources based on changing workload conditions
- A testing protocol that identifies hidden defects by applying a particular sequence of loads and measurements
- A manufacturing method that controls curing temperature and timing across defined stages to produce a more uniform material
Machines and devices
This includes physical devices whose components work together to perform a function. The invention may be an entirely new machine or an improvement to an existing one. Examples may include:
- A robotic joint that achieves a wider range of motion through a new gear arrangement
- A medical device with a redesigned fluid pathway that reduces blockages
- A battery-cooling assembly that directs airflow through a particular arrangement of channels to reduce temperature variation between cells
Articles of manufacture
This includes physical objects that don't have moving parts but are still made through a manufacturing process. Packaging materials, specialized tools, composite structures, coatings, and fabricated components can all qualify. Examples may include:
- A layered battery electrode structure that increases charge capacity
- A composite panel designed to improve strength without adding weight
- A bracket or fastener with a new geometry that makes installation more reliable
Compositions of matter
This includes new chemical compounds, mixtures, alloys, polymers, biological materials, and formulations. Examples may include:
- A battery electrolyte formulated to improve performance at low temperatures
- A polymer blend that combines flexibility with greater heat resistance
- A pharmaceutical formulation that improves the stability or delivery of an active ingredient
What kinds of inventions cannot be patented?
Most companies building physical products or chemical formulations will recognize their inventions in the categories above without much difficulty. The harder cases tend to involve software and process inventions — where the line between a patentable technical contribution and an unpatentable abstract idea is less obvious.
The distinction between patentable and non-patentable subject matter is often a question of how the invention is described, not just what it does.
Laws of nature and natural phenomena
You can't patent gravity, electromagnetism, or photosynthesis. You also can't patent the discovery of a naturally occurring substance, even if no one knew about it before. The Supreme Court ruled in Mayo Collaborative Services v. Prometheus Laboratories (2012) that a natural relationship between drug dosage and metabolite concentration was not patentable. Other examples include:
- A naturally occurring gene and its association with a disease
- The discovery that a particular biomarker predicts a medical outcome
- A mathematical relationship describing a physical process
A specific application of a natural phenomenon may still be patentable. The property that heat causes a polymer to change shape is not patentable by itself, but a medical stent designed to use that property in a specific, novel way may be.
General goals without a specific solution
A desired outcome is not an invention unless the application explains how the outcome is achieved. Examples include:
- "Use machine learning to improve customer engagement."
- "Automatically detect fraud."
- "Make a computer network more secure."
- "Reduce energy consumption in a manufacturing facility."
Each statement identifies a goal, but none describes the technical mechanism used to reach it. A patentable version would need to explain the particular system, method, or architecture responsible for producing the result.
Abstract ideas
This category has been the source of most patent eligibility disputes over the past decade, particularly for software. After the Supreme Court's Alice decision in 2014, examiners scrutinize whether a software application does more than implement an abstract idea on conventional computing hardware.
Abstract ideas include mathematical concepts, methods of organizing human activity (like business processes, contract structures, and social activities), and mental processes (things a human could theoretically perform without a computer). Examples may include:
- A new pricing model implemented through ordinary software
- A method for matching buyers and sellers using conventional database functions
- A contract structure carried out through a web platform
- A mathematical formula claimed without a specific technical application
For software companies, the central question is often whether the invention improves the operation of a computer or another technical system. A business rule that happens to run on software faces a much harder eligibility analysis than an algorithm that reduces memory usage or processing time in a technically specific way.
Creative expression
Patents do not protect the expression of an idea in artistic or literary form. These works are generally protected through copyright. Examples include:
- A novel, screenplay, or article
- A photograph or illustration
- A song or recorded performance
- The visual design of a website considered only as creative expression
In some cases, the ornamental appearance of a manufactured product may qualify for a design patent — a distinct category that protects how something looks rather than how it works.
Previously disclosed or publicly known inventions
An invention may lack novelty if an earlier patent, publication, public presentation, or publicly available product discloses all elements of the claimed invention. This applies regardless of whether the earlier disclosure came from your team or someone else.
The U.S. has a one-year grace period for an inventor's own disclosures, meaning you can still file up to 12 months after you publicly describe your invention. Most other countries don't offer this grace period, so an international filing strategy may require filing before any public disclosure.
Obvious variations of existing technology
An invention may also be rejected when the differences from existing technology would have been obvious to a skilled practitioner in the field. Examples might include:
- Replacing one conventional material with another known substitute without producing an unexpected result
- Combining familiar components that continue to perform their ordinary functions
- Adjusting a known operating range through routine testing
- Automating a manual process using standard technology without changing how the underlying process works
The invention does not need to be revolutionary. It does, however, need to involve more than the predictable use of existing techniques.
The gray area of patentability: framing matters
Some inventions sit near the boundary, and the way the application describes and claims the technical contribution can materially affect the eligibility analysis.
Take a recommendation engine. "A method for recommending products based on user preferences" may amount to an abstract business goal carried out with conventional software. A system that changes how a server processes requests by restructuring the order and weighting of feature evaluation — reducing latency in a technically specific way — may be patentable.
This is one reason careful patent drafting matters. An application should explain not only what the technology accomplishes, but also the technical mechanism responsible for the improvement and how that mechanism differs from what already exists.
Also, more detail does not automatically make an abstract idea patentable. The important question is whether the application describes a specific technological contribution rather than merely attaching a technical label to a business method.
How to recognize patentable inventions inside your company
Most founding teams undercount their patentable inventions. Some are visible in the features customers use, while others lie in the engineering decisions, internal processes, and supporting systems that make the product work.
Step 1: Look beneath important product features
A valuable product feature may itself be patentable, but its strongest patentable aspects often lie in the technical problems the team had to solve to make it work. For a real-time collaboration feature, the user-facing capability may be obvious, but the specific method for resolving conflicting edits across distributed nodes without data loss may not be.
Start with an important feature or capability and ask: What technical problems did we have to solve to build this? What did the standard approaches fail to handle? What did the team ultimately design, and how does it differ from conventional solutions?
Step 2: Identify decisions that weren't obvious
Patentable inventions often emerge from situations where the team tried the standard approach, found it didn't work, and had to develop something different. That "something different" may contain the invention.
Look for patterns like:
- "We couldn't use the standard library because it didn't handle our edge case, so we built a custom implementation that works by..."
- "The existing approach would have been too slow / too expensive / too unreliable, so we designed a different architecture that..."
- "Everyone in our space does it this way, but we found that doing it differently gives us a measurable advantage because..."
Step 3: Check for technical specificity
A patent application generally must describe the invention in enough detail that a person skilled in the relevant field could make and use it. The invention therefore needs to be developed beyond a design goal or product vision.
That does not necessarily mean you need a finished commercial product or working prototype before filing. It means the application must contain enough concrete detail to support and enable the invention as claimed — typically the mechanism, the architecture, or the sequence of steps that produces the improvement.
Step 4: Don't overlook process innovations
Patentable inventions aren't limited to products. If your company has developed a specific method for testing, manufacturing, calibrating, deploying, or operating something — and that method produces a better result than what was previously possible — it may be worth evaluating for patent protection.
Manufacturing companies frequently miss process patents. A specific sequence of steps that reduces waste, improves yield, or achieves tighter tolerances can be just as patentable as the product those steps produce.
Step 5: Check the disclosure timeline
Before diving too deep into the patent process, check whether the invention has already been publicly disclosed. Blog posts, conference presentations, published whitepapers, open-source code, public product launches, and investor materials can all trigger the disclosure clock.
If you're past the 12-month window on a particular invention, your U.S. filing rights on the disclosed material are likely gone. If you're within the window, a provisional patent application is the fastest way to establish a priority date while you develop the full application.
Not sure which of your ideas are worth filing on?
Patentext's platform helps companies identify, capture, and evaluate patentable work before it becomes a filing decision. Instead of guessing, you can run your engineering team's ideas through a structured evaluation that surfaces what matters and when.
Common mistakes companies make when evaluating patentability
Treating commercial value as proof of patentability
A feature can be highly valuable to customers without being patentable. Commercial value depends on whether the feature solves an important user problem. Patentability depends on whether the company can identify and describe the specific technical contribution responsible for making it work.
The better question is not simply, "Does this feature matter?" It is, "What did we have to invent in order to make it work?"
Evaluating inventions after they have been disclosed
Many companies do not decide against filing. They simply wait until after the invention has appeared in a product release, conference presentation, technical paper, or public demo.
At that point, the company may already have lost rights outside the United States, and the one-year U.S. grace period may be running. Patentability should therefore be considered before technical details become public — ideally as part of the product development process, not after the launch.
Looking only for major breakthroughs
Patentable inventions do not need to represent a new scientific field or an entirely new category of product. Many patents cover focused improvements to existing technology.
A change may be worth evaluating when it produces a meaningful technical benefit through a new design, method, or configuration. The relevant question is whether the improvement would have been obvious to a skilled practitioner given what already existed — not whether it represents a breakthrough.
A strong patent portfolio may include both major breakthroughs and smaller technical insights that collectively protect the advantages customers value and competitors may try to replicate.
Relying on a high-level product description
Founders and executives often describe what a feature does. Patent evaluation usually requires a closer explanation of how it works.
For example, "We use AI to automate compliance review" says very little about the underlying invention. The patentable contribution may lie in how the system extracts requirements, resolves conflicting rules, or surfaces relevant precedents — not in the general goal of automation.
The inventors who built the system should be part of the conversation because they can explain the technical decisions that distinguish it from a general product concept.
Limiting the search to customer-facing features
Patentable work commonly exists outside the core product. Internal systems can contain technical solutions that are just as valuable and sometimes harder for competitors to detect.
A company may have developed a better deployment process, a more reliable testing method, specialized manufacturing equipment, or an improved calibration workflow. Patent evaluation should cover the full scope of technical work, not just the features customers see.
However, an invention that remains entirely internal and would be difficult for competitors to reverse engineer may ultimately be better protected as a trade secret. The company should still identify it — the decision to patent versus maintain as a trade secret is itself a strategic one.
When should you evaluate an invention for patent protection?
Not every patentable invention is worth pursuing. A strong candidate generally combines a defensible technical contribution with meaningful business value: it may protect an important product advantage, give the company leverage in a competitive market, support a fundraising or licensing narrative, or establish priority over a direction competitors are moving toward.
Filing also requires time, money, and enough disclosure to explain how the invention works. The goal of an early evaluation is therefore to identify which technical advances justify that investment, while also flagging which inventions need to be documented before any public disclosure occurs.
And keep in mind that some inventions are better protected as trade secrets. This may be the stronger option when the technology is difficult to reverse engineer, can be kept confidential, and is likely to remain valuable over a long period of time. A structured evaluation process helps the company make that call deliberately rather than by default.
An invention may be worth evaluating when:
- Your team has developed a specific technical solution that appears meaningfully different from existing approaches
- A product launch, conference presentation, technical article, or open-source release could reveal how the solution works
- The invention supports an important product advantage that competitors may try to replicate
- You are building an IP portfolio for fundraising, licensing, partnerships, acquisition discussions, or entry into a competitive market
- Similar products or patents are beginning to appear, making it important to document what your team developed and when
This evaluation does not need to begin with a meeting or a fully formed invention disclosure. A structured process can help the team capture the problem, the technical solution, the alternatives considered, and the timeline — creating a record that supports both the patent filing and the priority date, if the company later decides to file.
Once the company decides to file, a registered patent agent or patent attorney can prepare the application, develop the claims, and handle prosecution before the USPTO. Patentext guides teams through this process from early invention capture through professionally prepared applications.
How does a company build a repeatable patentability evaluation process?
Treating invention identification as a one-time exercise misses the point. Companies that get meaningful value from their patent portfolios build a lightweight, repeatable process for surfacing and evaluating inventions on a regular basis — not just before a major product launch.
That process doesn't require a patent attorney on staff or having a patent agency on retainer. It just requires a structured way to capture technical problems and solutions from the people doing the work, before those solutions become public.
This is the problem Patentext is built to solve. The platform helps companies capture invention context from engineering teams, run structured evaluations, track decisions, and move approved inventions toward professionally prepared patent applications. See how the platform works. View pricing.
This article is for informational purposes only and does not constitute legal advice. Patent laws are complex and vary by jurisdiction. For personalized guidance, consult a qualified patent attorney or registered patent agent.
