Checklist · Intellectual Property

Drafting an Invention Disclosure: A Practical Checklist

Type
Checklist
Jurisdiction
National
Date
2025-08-23
Topic
Intellectual Property

An invention disclosure is the source document from which a patent is built, and its depth sets the ceiling on the protection an inventor can ever obtain. This checklist walks an inventor or R&D team through gathering exactly what patent counsel needs: a quantified problem statement, the mechanism rather than the result, the alternative embodiments you never built, the prior art you already know, the dated records that anchor conception, and the contributor roles that decide inventorship. It flags the timing bars under 35 U.S.C. § 102 (including Helsinn's secret-sale rule and Metallizing's forfeiture doctrine), the duty of candor under 37 C.F.R. § 1.56, and the § 112 written-description, enablement, and best-mode demands the disclosure must eventually satisfy. Worked hypotheticals show how each doctrine turns a routine engineering note into either a broad, defensible patent or a narrow one a rival designs around at lunch. Use it to produce a record that lets counsel claim the full invention and survive both examination and later attack.

Drafting an Invention Disclosure: A Practical Checklist

Intellectual Property > Patent | Published 23 August 2025 - Updated 9 September 2025 | UpLaw - uplaw.us

A patent can protect only what its application disclosed, and the application can disclose only what the inventor first wrote down. The invention disclosure is that first writing. This checklist turns a deceptively simple intake form into a doctrinally complete record — one built to satisfy §§ 101, 102, 103, and 112, honor the duty of candor, and give counsel room to claim the whole invention rather than the one version that happened to get built.

Keywords: invention disclosure, patent application, prior art, claim scope, alternative embodiments, conception, reduction to practice, inventorship, Pannu, duty of candor, 37 C.F.R. § 1.56, on-sale bar, public use, Helsinn, Metallizing, best mode, 35 U.S.C. § 112, written description, enablement

The disclosure is the source code of the patent

Think of the patent application as a compiled program and the invention disclosure as the source. A lawyer cannot claim a feature you never described, cannot enable a mechanism you never explained, and cannot distinguish prior art you never mentioned. Everything downstream — the claims that define the monopoly, the specification that supports them, the arguments that beat rejections, the record that survives a validity attack a decade later — is compiled from the raw material you hand counsel at the start. Feed in a thin, results-only summary and you get a thin, results-only patent: the kind a competitor engineers around over a long lunch. Feed in a deep, mechanism-rich record with a mapped range of alternatives and you give counsel the material to fence off a market for the full twenty-year term.

That is why the disclosure deserves more care than most inventors give it. The instinct after a breakthrough is to describe the thing you built — the exact prototype, the exact parameters, the exact chip. Patent law rewards the opposite instinct: describe the idea behind the prototype broadly enough that the claims reach every reasonable variation, yet concretely enough that a skilled reader could actually build them. Holding both of those in tension is the whole art of a good disclosure, and this checklist is organized around getting each half right.

A word on how to use what follows. Each phase pairs a set of concrete tasks — the checklist items you actually complete — with a short "why it matters" note tying the task to the doctrine it serves and the case law that will judge it. Work through the phases before your intake meeting with counsel; complete as many items as you can with counsel; and mark the finished document "CONFIDENTIAL — ATTORNEY-CLIENT PRIVILEGED" the moment you start it. Prepared for your lawyer, it is a privileged communication, and treating it that way from the first keystroke keeps it that way. If you want the narrative long-form treatment of the same material, pair this checklist with the companion guide on drafting an invention disclosure that maximizes patent scope and the foundational overview of how U.S. patents work.

What the disclosure must eventually satisfy

Before the tasks, the targets. A disclosure is good or bad only relative to the legal tests it is trying to feed. Five bodies of law sit at the end of the pipeline, and a disclosure written in ignorance of them tends to fail silently — nothing looks wrong until a rejection or a defendant proves otherwise.

Keep those five in the back of your mind as you read. Every phase below exists to put ammunition in one or more of them. For the doctrinal map in full, see the overview of the U.S. patent system and the plain-English FAQ for inventors and founders.

Phase 1 — Lay the administrative foundation

  • [ ] Give the invention a plain working title and a disclosure date.
  • [ ] List every contributor by name, with role, department or affiliation, contact information, residence, and citizenship. (Residence and citizenship still matter for some foreign filings and export-control screens.)
  • [ ] Identify the project, program, or grant behind the work, and every funding source.
  • [ ] Flag any federal funding by name and award number.
  • [ ] Mark the document "CONFIDENTIAL — ATTORNEY-CLIENT PRIVILEGED COMMUNICATION" and restrict distribution.
  • [ ] Put any deadline pressure — an imminent paper, demo, sale, or launch — at the very top, in bold, so counsel can triage timing before anything else.

Why it matters. People leave. Memories blur. The lab notebook gets boxed and shipped to storage. Your disclosure is a time capsule that a litigator you have never met may pry open in 2034, and the administrative header is what lets that stranger find the contributors and their records when it counts. The funding question is not boilerplate: research supported by federal money triggers the Bayh-Dole Act, 35 U.S.C. §§ 200–212, which imposes election, disclosure, and reporting obligations and reserves the government a nonexclusive license plus march-in rights under § 203. Miss the Bayh-Dole disclosure window and the funding agency can take title. Grant-funded university and startup inventors trip on this constantly, which is one reason IP hygiene belongs in the earliest founder tasks — see the startup incorporation checklist and the guide to getting diligence-ready before you raise.

A note on the word contributor. It is deliberately broader than inventor. At this stage you are casting a wide net over everyone who discovered, developed, tested, or analyzed any aspect of the invention. Who among them is a legal inventor is a determination for counsel in Phase 7 — do not prejudge it by leaving anyone off the header.

Phase 2 — State the problem before the solution

  • [ ] Write down the problem the invention solves before you describe the solution.
  • [ ] Quantify it. "The sensor drifts" is useless; "the MEMS gyroscope accumulates 4°/hour of bias drift because thermal gradients across the die shift the resonant frequency" is claimable.
  • [ ] Identify the root cause, not just the symptom.
  • [ ] Note whether the problem was long recognized and stubbornly unsolved, and whether others tried and failed — with names and dates if you have them.

Why it matters. A sharp problem statement does real legal work on non-obviousness under § 103. The Graham framework treats objective evidence — the "secondary considerations" — as part of the obviousness inquiry, not an afterthought, and a long-felt but unmet need together with the failure of others are two of the most persuasive items on that list. KSR loosened the rigid teaching-suggestion-motivation test and made it easier to call things obvious, which paradoxically raised the value of this objective evidence: real-world proof that skilled engineers wanted the result and could not reach it is a powerful answer to a hindsight-driven rejection. Counsel can only make that argument if you gave them the raw facts. See the deep treatment in obviousness under Section 103 and the tactical checklist for defeating a Section 103 rejection.

Worked example. Nimbus Robotics, a hypothetical drone maker, builds a delivery quadcopter whose flight time collapses in cold weather. The engineer who fixes it writes the disclosure two ways. Version A says: "We improved cold-weather battery performance." Version B says: "Below 5°C, lithium-ion cells lose roughly 30% of usable capacity because lithium plating on the graphite anode raises internal resistance; three prior thermal-wrap approaches added mass that cut payload and never recovered more than 10% of the lost range." Version B hands counsel a quantified problem, a named root cause, and a documented record that others tried and failed. Version A hands counsel nothing but a slogan. The claims that flow from B can be broad and still survive an obviousness challenge; the claims from A cannot even be drafted.

Phase 3 — Describe the mechanism, not the result

  • [ ] For each significant aspect, answer both how it works and why it works better than the alternatives.
  • [ ] Carry enough technical detail that a skilled person could build it: for an algorithm, the ordered steps and data structures; for a structure, the parts and their arrangement; for a process, the sequence and conditions.
  • [ ] Connect every technical choice to the benefit it produces in an explicit cause-and-effect chain.
  • [ ] Separate the essential features (the floor of your claim scope) from the optional ones.

Why it matters. This is the raw material for enablement under § 112(a). Enablement asks whether the specification teaches a person of ordinary skill to make and use the invention without undue experimentation — a standard the Federal Circuit measures against the eight non-exclusive factors of In re Wands, 858 F.2d 731 (Fed. Cir. 1988), including the quantity of experimentation needed, the amount of guidance provided, the presence of working examples, and the predictability of the art. A disclosure that recites results but hides mechanism forces counsel to choose between claims so narrow they only cover the prototype and claims so broad they are not enabled. Neither is what you want.

A trap worth naming. Discuss the invention's limits and failures out loud with counsel rather than memorializing colorful defeat language in the written record. A line like "this completely falls apart above 60°C" reads fine in a Slack message to a teammate and reads like a validity concession when a defendant's lawyer reads it aloud to a jury, stripped of context. The rule of thumb: tell your lawyer everything; let your lawyer decide what enters the permanent file. The teaching function of the patent system — the idea that the public gets a working instruction manual in exchange for the monopoly — is exactly what enablement enforces. See Sean B. Seymore, The Teaching Function of Patents, 85 Notre Dame L. Rev. 621 (2010).

Phase 4 — Map the alternatives you never built

  • [ ] For each aspect, name the broadest category — the genus — that captures your approach.
  • [ ] List specific examples within it — the species — flag your preferred one, and explain the tradeoffs across the range.
  • [ ] Capture the four families of alternative: component substitutions, process variations, configuration options, and parameter ranges (with preferred values inside each range).
  • [ ] For each genus, supply the bridges between species — design rules, selection criteria, structure-function relationships, worked examples — so a broad claim is genuinely supported, not merely aspirational.

Why it matters. Two doctrines pull in opposite directions here, and the disclosure has to satisfy both. On one side, you do not have to build the alternatives. A sufficiently detailed, enabling description is a constructive reduction to practice, and an invention can be "ready for patenting" on drawings and descriptions alone. Pfaff v. Wells Electronics, Inc., 525 U.S. 55, 67–68 (1998). So describing five variants you never prototyped is legitimate and valuable. On the other side sit the written-description and enablement limits of § 112. Amgen v. Sanofi is the modern warning shot: the Supreme Court held that a claim to an entire genus of antibodies defined by function was not enabled where the patent taught only a fraction of the members and left the rest to trial and error, reasoning that "the more a party claims, the broader the monopoly it demands, the more it must enable." 598 U.S. at 615. Written description asks the parallel question — whether the disclosure shows you possessed the full claimed genus, not just one member of it. Ariad, 598 F.3d at 1351.

The practical lesson: a single species does not automatically support a broad genus claim. It supports one only if you supply enough connective tissue — the rules that tell a skilled reader how to get from your one example to the others without undue experimentation. That connective tissue is precisely what Phase 4 asks you to write down.

Worked example. An inventor at a hypothetical materials startup discovers that swapping a particular ceramic filler into a battery separator suppresses the cold-weather resistance spike from Phase 2. The thin disclosure names the one filler she tested — call it Filler X. The strong disclosure names the genus ("metal-oxide fillers with a surface hydroxyl density above a threshold that scavenge trace water"), lists Filler X plus four untested candidates in the same class, states the measurable property that makes any member work, and gives a selection rule for choosing among them. If a rival later uses Filler Y — a cousin of X that the thin disclosure never contemplated — the broad claim reads on Filler Y and the narrow one does not. Same invention, wildly different fence. And because the strong version teaches why the genus works, it has a fighting chance of surviving the Amgen enablement question that would sink a bare "any metal-oxide filler" claim.

This scope map is not just a prosecution asset. It is the same document counsel will consult when running a later freedom-to-operate analysis on the product line, because the alternatives you claimed are the alternatives you are free to practice.

Phase 5 — Surface the prior art you already know

  • [ ] List every reference you are aware of — patents, published applications, journal and conference papers, theses, standards, datasheets, products on the market, public demos, YouTube videos — with enough detail (number, author, title, link) for counsel to pull it.
  • [ ] For each reference, note in a sentence how the invention differs.
  • [ ] Flag the areas where prior art probably exists but you have not searched.
  • [ ] Include your own earlier work — your prior publications, talks, and product launches, and your employer's prior sales.
  • [ ] Ask the honest question: "What did I read, see, or buy that pointed me toward this idea?" Then write all of it down.
  • [ ] Do not editorialize about patentability. List what you know and let counsel decide what to cite in an Information Disclosure Statement.

Why it matters. The duty of candor, 37 C.F.R. § 1.56, obliges every person substantively involved in the prosecution — inventors, in-house liaisons, prosecuting attorneys — to disclose information material to patentability. The mechanism for satisfying it is the Information Disclosure Statement under 37 C.F.R. §§ 1.97–1.98, and its timing rules are unforgiving. The reason the stakes run so high is the inequitable-conduct defense. In Therasense, the en banc Federal Circuit tightened the doctrine, holding that a challenger must ordinarily prove but-for materiality — that the PTO would not have allowed the claim had it known the reference — and a specific intent to deceive, with deceptive intent being "the single most reasonable inference" from the evidence. 649 F.3d at 1290–91. Tight as that standard is, the consequence of losing is nuclear: the entire patent is rendered unenforceable, and the taint can spread to related patents. Because the charge must be pleaded with particularity under Rule 9(b), the pleading gate of Exergen Corp. v. Wal-Mart Stores, Inc., 575 F.3d 1312 (Fed. Cir. 2009), filters weak claims early — but a genuinely hidden, material reference clears that gate easily.

The candor duty is not a reason to bury counsel in noise; over-disclosure has its own costs. It is a reason to tell counsel everything you know and let counsel exercise judgment about what is material. Your job on the disclosure is completeness, not curation. For the compliance mechanics, see the duty-of-candor checklist for avoiding inequitable conduct; for the litigation-side view, see inequitable conduct and the duty of candor and the deep dive on proving inequitable conduct.

Worked example. A software engineer building a fraud-detection model recalls that a graduate-school classmate presented a strikingly similar ensemble method at a workshop three years ago, and that an open-source library implements a nearby technique. He is tempted to leave both off the disclosure — surely his version is different enough. That temptation is the exact shape of an inequitable-conduct case. He should list the workshop talk, the library, and a one-line note on how his approach differs, then let counsel decide whether either is material. Disclosed and distinguished, the references are harmless and may even strengthen the file. Concealed and later unearthed in discovery, they are the raw material for an argument that the whole patent should fall.

Phase 6 — Anchor conception, diligence, and reduction to practice

  • [ ] Record the conception date — the moment a definite and permanent idea of the complete and operative invention formed — and the notebook page, commit hash, or file that anchors it.
  • [ ] Record the development milestones: first working example, first successful test, first disclosure to another person (and to whom).
  • [ ] Confirm notebook hygiene: entries signed and dated, witnessed near the time of the work by a qualified non-inventor, and failures recorded alongside successes.
  • [ ] Cross-reference related notebooks and explain any gaps in the dates.
  • [ ] Keep legal characterizations out of the notebook — no musings about patentability, no disparagement of competitors, no transcription of conversations with counsel.

Why it matters. Conception is the legal heart of invention. The classic formulation from Burroughs Wellcome Co. v. Barr Laboratories, Inc., 40 F.3d 1223, 1228 (Fed. Cir. 1994), is that conception is complete "when the inventor has formed in his or her mind a definite and permanent idea of the complete and operative invention, as it is thereafter to be applied in practice" — clear enough that only ordinary skill, not further invention, is needed to build it. Reduction to practice comes in two flavors: actual (you built it and it worked for its intended purpose) and constructive (you filed an enabling application). The America Invents Act moved the United States to a first-inventor-to-file system, which demoted the old priority contests, but conception and diligence still surface constantly — in derivation proceedings, in swearing behind certain references, in inventorship disputes, and in litigation over who invented what.

Here is the hard-won truth about records: uncorroborated inventor testimony about conception is treated with deep suspicion. Courts have distrusted the lone inventor's after-the-fact say-so since at least The Barbed Wire Patent, 143 U.S. 275 (1892), where the Supreme Court discounted oral recollections offered years later to antedate a patent. A notebook that is signed, dated, and witnessed by a non-inventor who actually watched the work turns that weak testimony into strong corroboration; an unsigned, unwitnessed file gets no more weight than the testimony it was supposed to support. Good notebook discipline is also the backbone of any trade secret protection program, so the habit pays twice.

Worked example. Two hypothetical co-founders sketch a novel data-compression scheme on a whiteboard on a Friday. One photographs the board, emails it to both of them (timestamping it), and logs a witnessed notebook entry the following Monday describing the algorithm in enough detail that a competent engineer could implement it. The other keeps the idea in his head and codes it up months later. If a priority or derivation question ever arises, the first founder's corroborated, dated record is worth a great deal and the second founder's memory is worth almost nothing. The invention was the same on Friday; only the record differs.

Phase 7 — Capture contributor roles for inventorship

  • [ ] Document, in detail, what each contributor actually did, cross-referenced to dates and records.
  • [ ] Identify everyone who created, discovered, developed, tested, or analyzed any aspect — inside or outside the organization.
  • [ ] Do not pronounce on inventorship. Give counsel the raw facts and let counsel make the legal call.
  • [ ] Designate one inventor to coordinate and serve as counsel's primary contact.
  • [ ] Have every potential inventor review the disclosure for accuracy before it is submitted.
  • [ ] Flag any meaningful role an AI system played in conceiving a feature — do not paper over it.

Why it matters. Ownership flows from inventorship: absent an assignment, patent rights vest initially in the inventors, a default the Supreme Court reaffirmed in Board of Trustees of the Leland Stanford Junior University v. Roche Molecular Systems, Inc., 563 U.S. 776 (2011). Get the named inventors wrong and you cloud title, jeopardize standing to sue, and — if the error was deceptive — hand a defendant an invalidity argument. The governing test comes from Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998): a joint inventor must (1) contribute in some significant manner to the conception of the invention, (2) make a contribution that is not insignificant in quality when measured against the full invention, and (3) do more than merely explain well-known concepts or the current state of the art. Under that test, the venture capitalist who funded the lab, the manager who set the goal, and the technician who ran the assay exactly as instructed are usually not inventors, while the summer intern who suggested the one insight that made everything click may well be. An innocent naming error is fixable under 35 U.S.C. § 256; a deceptive one is a validity problem. See the joint-inventor rules of § 116 for how multiple inventors are handled on a single application.

Because inventorship decides who invented and assignment decides who owns, the disclosure feeds directly into the company's chain-of-title paperwork. The two must line up, which is why employee invention assignment agreements should be in place before the disclosure is written, not scrambled for afterward.

The AI wrinkle. If a generative model did more than autocomplete — if it proposed the specific structure, parameter, or step that solved the problem — say so plainly. Under Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022), an "inventor" under the Patent Act must be a natural person, so an AI system cannot be named. But the invention is not automatically unpatentable; the USPTO's Inventorship Guidance for AI-Assisted Inventions, 89 Fed. Reg. 10,043 (Feb. 13, 2024), applies the Pannu factors to ask whether a natural person made a significant contribution to conception notwithstanding the machine's role. Concealing a material AI contribution invites both an inventorship defect and a candor problem. The frontier here is genuinely unsettled — for the full picture see who owns AI-generated inventions and the cross-jurisdictional survey in AI and inventorship.

Worked example. A hypothetical pharma team runs a generative chemistry model that outputs a shortlist of candidate molecules; a staff chemist recognizes why one candidate should bind a difficult target, refines it, and confirms activity in the lab. The disclosure should describe the model's role honestly and detail the chemist's contribution — the recognition, the refinement, the reduction to practice. Named correctly, the chemist is the inventor and the patent is sound. Hidden, the AI's role becomes a latent defect that surfaces at the worst possible moment.

Phase 8 — Catalog the timing bars and public disclosures

  • [ ] List every public disclosure: publication, conference talk, posted video, live demonstration, thesis deposit, beta deployment — with dates and audiences.
  • [ ] List every commercial event: offers for sale, actual sales, price quotes sent to customers, distributor and beta agreements — with dates and whether under NDA.
  • [ ] If a potentially barring event has already happened, calendar the one-year U.S. grace-period deadline immediately and confirm any filing beats it.
  • [ ] If foreign protection is on the table, plan to file before any public disclosure — most of the world demands absolute novelty and grants no grace period.

Why it matters. Section 102(a)(1) makes prior art of anything "patented, described in a printed publication, or in public use, on sale, or otherwise available to the public" before the effective filing date — anywhere in the world. The § 102(b)(1) grace period forgives the inventor's own disclosures made within one year of filing, but that grace period is largely an American indulgence; rely on it and you forfeit most foreign rights. Three lines of authority make this phase more dangerous than it looks:

  • The on-sale bar is triggered by commercial character, not publicity. Pfaff set the two-part test: the invention must be (1) the subject of a commercial offer for sale and (2) ready for patenting. Courts read "commercial offer" against ordinary UCC principles, so a routine quote or purchase order can qualify. In Helsinn Healthcare S.A. v. Teva Pharmaceuticals USA, Inc., 586 U.S. 123 (2019), the Supreme Court held that even a secret, confidential commercial sale — a supply agreement whose terms were never made public — triggers the AIA on-sale bar. What looks to a founder like a quiet, NDA-protected business deal can look to a litigator like a statutory bar with a one-year fuse.
  • Public use can be startlingly private. Egbert v. Lippmann, 104 U.S. 333 (1881) — the corset-spring case — held that a single unrestricted use by one person, out of public view, can still be a barring "public use." The mirror-image safe harbor is City of Elizabeth v. American Nicholson Pavement Co., 97 U.S. 126 (1877), where testing a wooden pavement on a public toll road for six years was not a public use because it was bona fide experimentation. Whether your field trial is "experimental" or "public" turns on control, records, and intent — all of which trace back to Phase 6.
  • Your own secret commercial use can forfeit the patent. Under Judge Learned Hand's forfeiture doctrine in Metallizing Engineering Co. v. Kenyon Bearing & Auto Parts Co., 153 F.2d 516 (2d Cir. 1946), an inventor who commercially exploits a secret process for more than the grace period before filing forfeits the right to patent it — you cannot have both trade-secret head-start profits and a later patent monopoly on the same process. That doctrine punishes exactly the "use it quietly, patent it later" instinct that startups often follow by default.

Worked example. A hardware startup, eager for revenue, emails a signed price quote for its still-secret sensor module to a prospective OEM customer fourteen months before filing a patent application, all under NDA. The founders assume secrecy preserves their rights. It does not. Under Pfaff and Helsinn, the confidential offer likely started the on-sale clock, and fourteen months is past the one-year grace period. A single sales email has quietly narrowed — possibly eliminated — the patent that would have protected the company's core product. The fix costs nothing if you calendar these events in real time and file first; it is unfixable after the clock runs.

Because these bars intersect directly with launch planning, the timing analysis should travel alongside the freedom-to-operate search checklist and the broader IP clearance toolkit.

Phase 9 — Best mode and the rest of the § 112 quartet

  • [ ] Identify the embodiment you currently consider best, and explain why — do not withhold the "secret sauce."
  • [ ] Confirm that any range or genus you describe is itself described and enabled across its whole scope, not just at the one value you happened to test.
  • [ ] Draft claims-in-waiting in your head: for each essential feature, ask whether the disclosure would let counsel claim it with a definite boundary.
  • [ ] Attach labeled drawings, flowcharts, schematics, or photographs — and make sure every reference numeral in the text appears in a figure.

Why it matters. Section 112(a) requires the specification to set forth the best mode contemplated by the inventor of carrying out the invention. The AIA removed best-mode failure as an invalidity or unenforceability defense, 35 U.S.C. § 282(b)(3)(A), which tempts some inventors to hold back their best implementation. Resist the temptation for two reasons. First, the best-mode requirement itself was not repealed — only the litigation remedy was narrowed — and concealment can still color an inequitable-conduct narrative. Second, and more practically, the material you would withhold is usually the very material that makes the enablement and written-description case. Enablement is judged against the full scope of the claims (Amgen), written description against whether you possessed the full invention (Ariad), and both are measured through the Wands undue-experimentation factors. A claim can also die at the other end of § 112: under § 112(b) and Nautilus, claims must inform a skilled reader "with reasonable certainty" of their scope, 572 U.S. at 910, and vague functional language that seemed harmless in the disclosure becomes an indefiniteness rejection at the PTO or an invalidity finding in court. The disclosure is where the definiteness problem is born or avoided. For how these rejections play out and how to answer them, see the checklist for responding to a patent office action and the guide on turning patent rejections into granted claims.

Phase 10 — Tailor the disclosure to the invention type

Different technologies fail § 112 and § 101 in different ways, so the disclosure should lean into whatever the examiner and a future defendant will attack.

  • [ ] Software and algorithms. Describe the system from the computer's point of view — architecture, modules, data structures, and the algorithm as ordered steps, with flowcharts or pseudocode — and frame the invention as a technical solution to a technical problem. Name any open-source components (they carry license and disclosure consequences). This framing is your defense against the abstract-idea trap; work it against the Section 101 eligibility checklist and the post-Alice strategy guide.
  • [ ] Mechanical and hardware. Provide multiple drawing views, preferred materials, critical dimensions and tolerances, and any novel manufacturing method. If appearance is commercially important, consider whether a design patent belongs alongside the utility filing — see the design-patent-or-utility-patent checklist.
  • [ ] Chemical and pharmaceutical. Give precise structures, synthesis routes, characterization data, and — crucially, after Amgen — which structural modifications preserve activity and which destroy it, so a genus claim has real support.
  • [ ] Business methods. Identify the specific technical systems that implement the method and articulate why it is more than an abstract idea running on generic hardware. This is where most business-method claims live or die under Alice.

Phase 11 — Decide the patent-versus-trade-secret fork honestly

  • [ ] Ask, before filing anything, whether this invention is better protected as a patent or as a trade secret.
  • [ ] Weigh detectability: if you cannot tell from a competitor's product whether they used your method, a patent may be unenforceable in practice and a secret may be worth more.
  • [ ] Weigh durability: patents expire in twenty years; a well-kept secret can last indefinitely — but evaporates the instant it leaks or is independently discovered.
  • [ ] Remember that publishing a patent application destroys trade-secret status in what it discloses, and that filing after long secret commercial use can trigger the Metallizing forfeiture from Phase 8.

Why it matters. The disclosure is the last comfortable moment to choose your regime, because the choice is partly irreversible. Patenting demands publication; trade-secret protection demands the opposite. Some inventions — a manufacturing process no teardown could reveal — are worth more as secrets; others — a visible product feature a rival can copy on sight — are worthless as secrets and must be patented to be protected at all. The two regimes can even be layered, patenting the detectable elements while keeping the undetectable know-how dark. For the framework, see trade secret law, the practical trade-secret audit checklist, and the side-by-side comparison in copyright, trademark, patent, and trade secret compared. For software specifically, the layering strategy is mapped in how the law protects software. Whatever you choose, protect the disclosure itself with a properly drafted confidentiality regime — the NDA drafting checklist and the guide to NDAs for technology deals cover the mechanics.

Phase 12 — Final review and submission

  • [ ] Confirm the disclosure captures the problem, the mechanism, the alternatives, the advantages, the prior art, the key dates, and the contributors.
  • [ ] Verify every potential inventor has reviewed and signed off on the accuracy of what is attributed to them.
  • [ ] Re-confirm the confidentiality marking and that the invention has not been publicly disclosed or offered for sale without counsel's sign-off.
  • [ ] Submit as soon as the invention is developed enough to describe in an enabling way — do not wait for an imminent launch to force the issue.

Why it matters. Timing failures are the most common way good inventions become unpatentable, and they are almost always avoidable. The single most valuable habit an R&D organization can build is to route disclosures to counsel early and routinely, not in a panic the week before a trade show. Everything upstream in this checklist is wasted if the filing lands a day after a bar date. The disclosure then flows into the full prosecution pipeline mapped in the patent prosecution roadmap and explained procedurally in USPTO patent procedure.

Common failure modes

The same handful of mistakes recur across industries. Each maps to a phase above.

  • Describing only what you built. No alternatives means a narrow, sidesteppable patent. (Phase 4.)
  • Omitting the problem. Starves the non-obviousness argument and leaves counsel with nothing to build a Graham record from. (Phase 2.)
  • Marketing language instead of technical language. "Revolutionary breakthrough" is not something a lawyer can claim or an examiner can allow. (Phase 3.)
  • Assuming counsel already knows your field. Undefined terms produce ambiguous claims and indefiniteness problems under Nautilus. (Phase 9.)
  • Hiding prior art. Risks unenforceability of the entire patent under Therasense. (Phase 5.)
  • Waiting too long. A publication, sale, or public use quietly ripens into a § 102 bar; a confidential quote can do it under Helsinn. (Phase 8.)
  • Getting inventorship wrong. Clouds ownership under Stanford v. Roche and, if deceptive, invalidates under Pannu. (Phase 7.)
  • Colorful failure language in the file. Handing a defendant a validity concession stripped of context. (Phase 3.)

Putting it together: one invention, two disclosures

To see how much the disclosure decides, run one hypothetical all the way through. A hypothetical medical-device engineer, Priya, invents a way to cut motion artifacts in a wearable ECG patch by adaptively filtering the signal based on an onboard accelerometer.

The weak disclosure is a page long. It says the patch "uses motion data to clean up the ECG signal," names the one microcontroller and the one filter she coded, attaches no drawings, lists no prior art because "ours is different," and says nothing about the three noisy prototypes that failed before this one worked. It does not mention that Priya demoed the working patch at a cardiology conference eleven months ago, or that her employer emailed a spec sheet to a hospital buyer. From this, counsel can draft a claim to a patch running that specific filter on that specific chip — a claim a competitor avoids by changing either one.

The strong disclosure states the problem quantitatively (motion artifacts corrupt ambulatory ECG readings above a defined activity threshold, defeating three prior filtering approaches she documents by name). It describes the mechanism as ordered steps, generalizes the filter to a genus of adaptive algorithms with a selection rule, names four accelerometer configurations she did not build, lists the conference demo and the spec-sheet email with exact dates, credits the summer intern who suggested the activity-threshold gating, and flags the accelerometer-fusion library she borrowed. Counsel reading this can draft broad genus claims with real § 112 support, calendar the eleven-month conference demo so the filing beats the § 102(b)(1) bar with a month to spare, name inventors correctly, and file an IDS that inoculates the patent against a candor attack.

Same invention. Same engineer. The difference between a patent worth licensing and a patent worth nothing is the document Priya wrote before she ever met her lawyer. For the strategic lessons that flow from disputes over exactly these choices, see landmark patent wars and pivotal patent decisions of late 2024.

Primary and secondary authority

Statutes and regulations. 35 U.S.C. § 101 (eligibility; inventorship); § 102 (novelty; on-sale and public-use bars; grace period); § 103 (non-obviousness); § 112(a)–(b) (written description, enablement, best mode, definiteness); §§ 115, 116, 256 (inventor's oath; joint inventors; correction of inventorship); § 282(b)(3)(A) (best mode not an invalidity defense); §§ 200–212 (Bayh-Dole; march-in rights under § 203). 37 C.F.R. § 1.56 (duty of disclosure); § 1.63 (inventor's oath or declaration); §§ 1.97–1.98 (Information Disclosure Statements).

Cases. Graham v. John Deere Co., 383 U.S. 1 (1966); KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007); Pfaff v. Wells Elec., Inc., 525 U.S. 55 (1998); Helsinn Healthcare S.A. v. Teva Pharms. USA, Inc., 586 U.S. 123 (2019); Metallizing Eng'g Co. v. Kenyon Bearing & Auto Parts Co., 153 F.2d 516 (2d Cir. 1946); Egbert v. Lippmann, 104 U.S. 333 (1881); City of Elizabeth v. Am. Nicholson Pavement Co., 97 U.S. 126 (1877); Pennock v. Dialogue, 27 U.S. (2 Pet.) 1 (1829); Pannu v. Iolab Corp., 155 F.3d 1344 (Fed. Cir. 1998); Burroughs Wellcome Co. v. Barr Labs., Inc., 40 F.3d 1223 (Fed. Cir. 1994); The Barbed Wire Patent, 143 U.S. 275 (1892); Therasense, Inc. v. Becton, Dickinson & Co., 649 F.3d 1276 (Fed. Cir. 2011) (en banc); Exergen Corp. v. Wal-Mart Stores, Inc., 575 F.3d 1312 (Fed. Cir. 2009); Amgen Inc. v. Sanofi, 598 U.S. 594 (2023); Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d 1336 (Fed. Cir. 2010) (en banc); In re Wands, 858 F.2d 731 (Fed. Cir. 1988); Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014); Alice Corp. v. CLS Bank Int'l, 573 U.S. 208 (2014); Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66 (2012); Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022); Bd. of Trustees of the Leland Stanford Junior Univ. v. Roche Molecular Sys., Inc., 563 U.S. 776 (2011).

Administrative and secondary. USPTO, Inventorship Guidance for AI-Assisted Inventions, 89 Fed. Reg. 10,043 (Feb. 13, 2024); U.S. Patent & Trademark Office, Manual of Patent Examining Procedure §§ 2138.04 (conception), 2133.03 (on-sale and public-use), 2001–2004 (duty of disclosure), 2161–2164 (written description and enablement) (9th ed.); Donald S. Chisum, Chisum on Patents (treatise); Robert P. Merges & John F. Duffy, Patent Law and Policy: Cases and Materials; Janice M. Mueller, Mueller on Patent Law; Sean B. Seymore, The Teaching Function of Patents, 85 Notre Dame L. Rev. 621 (2010). USPTO inventor resources: https://www.uspto.gov/patents/basics.

This checklist is general legal information, not legal advice. Patent law changes, and the application of these doctrines is intensely fact-specific. Confirm the current state of the law and consult a qualified patent attorney before acting.

Related UpLaw Resources

Related Guides

Related Checklists

Related Toolkits

Related Articles