So, Do You Really Want to File a Patent?

Lessons From Doing It Myself


I’ve been an inventor on several patents during my time at Microsoft and Clearbrief, but in those cases I had the luxury of patent attorneys handling the heavy lifting. I would describe the idea, write a technical explanation, and then the legal team would transform it into a polished application with claims, formatting, examiner correspondence — the whole package.

This time was different. I had an idea I believed was patent‑worthy, but I was completely on my own. No attorneys. No corporate infrastructure. Just me, a blank document, and two AI assistants — Copilot and Gemini — to help me figure out what the USPTO actually expects.

What follows is what I wish someone had told me before I started.


1. Validating the Idea (or: Asking AI if I’m Crazy)


I began with a short summary and a bulleted list of the core concepts. I asked both Copilot and Gemini whether the idea seemed original and whether it had potential value. Both said yes. That was enough encouragement to push forward.

This is the first trap: AI will almost always tell you your idea is novel. It’s not lying — it simply can’t perform a real prior‑art search. But at the time, I took the green light and kept going.


2. Understanding the Patent Structure (and Realizing I Didn’t)


I knew patents followed a standard format — you can see it in any published application — but the exact order and formatting rules weren’t obvious. Gemini became my primary drafting partner, with periodic checks using Copilot.

Together we worked on:
  • Summary of the invention
  • Detailed description
  • Claims
  • Abstract

Fun fact: the abstract goes at the end of the document when you file, not the beginning.

The claims were the hardest part by far. Claims are the legal definition of what you want to protect. They must follow a rigid structure: one or more independent claims (the core idea) and dependent claims (variations or narrower versions).

AI can help, but it tends to over‑generate. Which leads to the next problem.

 
Image generated by Gemini (Google) based on user prompt.


3. The $20,000 Surprise: Claim Fees


Neither AI system warned me that the USPTO’s basic filing fee only covers:
  • Up to 3 independent claims
  • Up to 20 total claims

During brainstorming, we enthusiastically generated over 100 claims. I was proud — until I tried to file and saw the fee estimate: over $20,000.

I went back and, with AI’s help, compressed everything down to 20 total claims. It was painful but necessary.

Lesson: AI will happily generate more claims than you can afford.


4. Drafting With AI: Helpful, but Needing Constant Steering


Gemini was extremely helpful for drafting, but it had a tendency to:
  • Expand sections too much
  • Repeat content
  • Add new ideas that drifted away from my core invention

This is another trap: AI wants to be helpful, and “helpful” often means “let me add more stuff.” But more stuff dilutes your invention and increases filing costs.

My advice: stick to your guns. Keep the invention tight. Don’t let the model wander.


5. Figures: Use the Minimum Required


The USPTO requires at least one figure. I made the mistake of adding several. In hindsight, one figure would have been enough.

Every figure must be:
  • Labeled
  • Described in the text
  • Cross‑referenced

More figures = more work and more opportunities for formatting errors.


6. Provisional vs. Non‑Provisional: Timing Matters


Before talking to anyone, I recommend emailing yourself a detailed description of the idea. It’s not strong legal protection, but it’s something.

If you need to discuss the idea externally, file a provisional application. It’s cheap, nobody reads it, and it locks in your filing date.

My mistake: I filed the provisional too early. Then I made major changes and had to file a second non‑provisional. You have one year from your provisional filing to submit the non‑provisional.


7. Filing With the USPTO: Simple… Until It Isn’t


Uploading documents through the USPTO website is straightforward — until you discover the hidden requirements.

For example: The system does not automatically generate the Oath/Declaration where you attest that you’re the inventor. I assumed it would. It didn’t.

A month later, during pre‑review (where they check formatting), I received a notice that I owed another $68 because the declaration was missing.

The formatting checks are also surprisingly shallow. The system won’t catch:
  • Incorrect paragraph spacing
  • Missing figure references
  • Inconsistent headings

You find out about these issues a month after filing, when the USPTO emails you a cryptic notice.

Correcting the mistakes is one thing. Understanding how to re‑file the corrected documents is another. Neither AI assistant had the procedural nuance to guide me through that part. After a long session with Gemini and a lot of reading, I eventually figured it out.

If you get stuck, there is a help email: HelpAAU@uspto.gov.


8. Prior Art: What AI Can (and Cannot) Do


Both Copilot and Gemini warned me early on that they cannot perform a real prior‑art search. They don’t have access to unpublished applications, and they can’t guarantee novelty. What they can do is help you think strategically.

They gave me:
  • search terms
  • synonyms
  • related concepts
  • recommended databases
  • strategies for broadening or narrowing the search

Using their suggestions, I searched Google Patents, USPTO, WIPO, and the broader web. I found things that looked uncomfortably close to my idea — the classic solo‑inventor panic moment.

But when I fed those results back into Gemini, it helped me compare and contrast them with my invention. Both AI systems agreed: the items I found were related, but not the same. They helped me articulate the differences clearly.

This process:
  • Calmed me down, because “similar” is not “identical.”
  • Strengthened my application, because I could emphasize the unique aspects of my invention.

AI can’t certify novelty, but it can help you understand the landscape and refine your claims.


9. Creating Figures: Keep It Simple, Use Draw.io, and Label Everything


For patent figures, the tool that saved me was Draw.io (https://app.diagrams.net/). It’s free, browser‑based, and perfect for clean, USPTO‑friendly diagrams. Even better, Gemini can generate Draw.io XML, which you can paste directly into the editor and refine visually.

A few practical lessons:
  • Use simple shapes — no color or shading needed.
  • Label every component you reference in the text.
  • Use 12‑point font or larger so the USPTO’s OCR can read your labels.
  • Stick to one figure unless absolutely necessary.

If you need help generating a starting point, AI can sketch a layout, generate XML, or suggest labeling conventions. Just remember: clarity beats beauty.


10. A Work in Progress: The Journey Isn’t Over


Everything I’ve described so far only covers the initial stages of filing. I haven’t cleared all the early hurdles yet, and the USPTO process moves at a glacial pace. After the pre‑exam processing and formatting checks, it will take over two years before an examiner actually reviews my application.

That means there are still many opportunities for surprises, mistakes, or new requirements I haven’t encountered yet.

This article is a snapshot of where I am right now — not the final word. As I move through the next phases, respond to office actions, and (hopefully) reach the examination stage, I expect to learn a lot more. Some of it may validate what I’ve written here; some of it may contradict it. Either way, I’ll keep updating this article so it reflects the real, unfiltered experience of filing a patent on your own.

The journey is long, and I’m still in the early chapters.


11. Acknowledging the Work of USPTO Staff


Before going any further, I want to recognize the people on the other side of this process — the USPTO staff and patent examiners. It’s easy to get frustrated with delays, cryptic notices, or formatting requirements, but the reality is that the USPTO handles an enormous volume of work. Every year, they receive hundreds of thousands of patent applications across every imaginable field of technology. Each examiner is responsible for a large docket, often spanning multiple technical domains, and they must review every claim, every figure, every reference, and every cited piece of prior art with legal precision.

The backlog isn’t a sign of inefficiency — it’s a reflection of the sheer scale and complexity of the system. Patent examiners are doing meticulous, high‑stakes work under intense volume pressure. Even the pre‑review staff who check formatting and completeness are juggling thousands of submissions. Despite the bottlenecks, every interaction I’ve had has been professional and helpful. The system may be slow, but the people inside it are doing difficult work that deserves recognition.


12. Understanding USPTO Fees: Micro Entity, Small Entity, Regular — and Page Count


One thing I wish I had understood earlier is how dramatically USPTO fees change depending on your entity status. The USPTO has three categories, and the difference between them can be thousands of dollars over the life of a patent.

Regular (Large) Entity


This is the default category. You are a large entity if you do not qualify for small or micro status.
Typical examples: corporations with more than 500 employees, or anyone assigning rights to such a company.

Regular fees are the baseline — the most expensive tier.

Small Entity (60% discount)


Most independent inventors fall into this category. You qualify as a small entity if:
You are an individual inventor, or
A small business with fewer than 500 employees, and
You have not assigned (and are not obligated to assign) the invention to a large company.

Small entity status cuts most fees by 60%.

Micro Entity (75% discount)


Micro entity is the most restrictive category. To qualify, you must meet all of the following:
Your gross income is below a specific threshold (about 3× the U.S. median household income).
You have filed no more than four U.S. non‑provisional patent applications in your lifetime.
You have not assigned (and are not obligated to assign) the invention to a large entity.
You do not have a prior obligation to assign the invention to an employer who is not a micro entity.

Most inventors do not qualify for micro entity unless they are students, first‑time filers, or low‑income inventors.

Why This Matters


The fee differences are not small. For example:
Filing, search, and examination fees for a small entity total around $1,160.
The same fees for a large entity are roughly $2,900.
Micro entity fees would be around $290.

Maintenance fees (due at 3.5, 7.5, and 11.5 years) also scale by entity status.

The Hidden Trap: Claim Fees

The base fee covers:
3 independent claims
20 total claims

Extra claims get expensive fast — especially for regular entities. This is why my initial 100‑claim brainstorm turned into a $20,000 shock.

Another Hidden Trap: Page Count Fees

The USPTO charges extra if your application exceeds 100 pages. The surcharge applies for every additional 50 pages or fraction thereof.

This means:
A 101‑page application triggers the fee.
A 151‑page application triggers it twice.
AI‑generated expansions, repeated text, or too many figures can quietly push you over the limit.

This is yet another reason to keep your document tight and aggressively trim unnecessary content.

Practical Advice

  • Determine your entity status before drafting claims.
  • Keep your claim count under control.
  • Watch your page count — it can quietly inflate your fees.
  • Re‑confirm your status when filing — the USPTO will not correct it for you.

Understanding these categories early can save you money, time, and a lot of stress.
 
 




Popular posts from this blog

The Renaissance We Owe Ourselves

Image
There are periods in history when culture seems to move with the force of weather—fronts of ideas rolling in, colliding, breaking apart, leaving behind new landscapes. The twentieth century was one of those meteorological eras. Modernism fractured the old world; jazz rewired rhythm; cinema taught us to dream in images; rock and hip‑hop remixed rebellion; digital art dissolved the boundaries of medium altogether. Each decade felt like a new continent had risen from the sea. Today, the horizon looks different. We live in an age of astonishing creative abundance—more books, more films, more songs, more images than any human could hope to absorb in a lifetime. And yet, strangely, much of it feels familiar, as if we are walking through a museum where the paintings keep rearranging themselves but never quite change. Novels echo the emotional scaffolding of earlier classics. Films recycle the same arcs, the same reversals, the same lines of dialogue that once felt clever and now feel like déj...
Read more

Rethinking Copyright in the Age of Algorithmic Creativity

Image
Copyright law was built for a world that no longer exists. When the first modern copyright statutes emerged in the eighteenth and nineteenth centuries, creative production was slow, scarce, and largely controlled by professional gatekeepers. Books required printing presses. Paintings took months to complete. Music circulated through physical scores and performances. In that environment, it made sense for originality to be judged by human experts—art historians, lawyers, and judges—who could manually compare works and decide whether one meaningfully borrowed from another. But today, this human-centered system is buckling under the weight of digital abundance. Millions of images, songs, and designs circulate daily, and the institutions meant to protect creators can no longer keep pace. The result is a copyright regime that is both overburdened and unevenly enforced. Large corporations can marshal legal teams to defend their claims; independent artists often cannot. Determining whether a ...
Read more

The Case for a National Mission

Image
Every human enterprise — a family, a company, a country — runs on some shared sense of purpose. When that purpose is obvious, no one needs to talk about it. When it isn’t, people start holding meetings. Anyone who has endured a corporate “mission‑statement rollout” knows the dread of that moment: the sudden realization that the organization has forgotten what it’s for and is now trying to remember by committee. A country is, in the end, a very large group of people attempting to move in roughly the same direction. It needs a reason to do so that’s bigger than paperwork or habit. When that reason is strong, disagreements behave themselves; they stay in the background, like the hum of a refrigerator. When the reason fades, the hum becomes the whole soundtrack. The result isn’t necessarily hatred — more often it’s a kind of national restlessness, a search for meaning that leaves people unusually susceptible to cynicism, tribalism, or the comforting simplicity of blaming one another. F...
Read more