Frequently Asked Questions

Traditionally, a law firm might charge $7,500 to $15,000 for a single application. By doing the initial heavy lifting yourself with AI and using our review service, inventors often save 50% to 75% on the initial filing costs while still receiving professional attorney representation. 

In simple terms, Sycophancy means AI models are trained to be helpful assistants and they can aim to please you rather than tell you the truth or give you bad news. This means that AI can encourage an inventor to file patent applications or patent strategies that are doomed from the start due to a propensity to please over objective assessments of patentable or strategic merit.

 Absolutely yes. Patent law and patent application filing are governed by federal law and administered by the U.S. Patent and Trademark Office (USPTO), not by individual states. This means any registered patent attorney in good standing with the USPTO can represent inventors nationwide in preparing, filing, and prosecuting patent applications, regardless of where the client or attorney is physically located.

 SOME CRITICAL PITFALLS include the following:

• Public Disclosure & Confidentiality Risks: Many popular AI tools use your prompts to train their models. If you input the "secret sauce" of your invention into a public AI, it could be legally considered a public disclosure, which may prevent you from ever obtaining a patent.
  
• The "Claim Scope" Trap: AI often struggles with the legal nuance of patent claims. It may produce claims that are too broad (leading to an immediate USPTO rejection) or too narrow (meaning competitors can easily "design around" your patent and steal your idea).
  
• Technical "Hallucinations": AI is known to invent facts, technical citations, or prior art that don't exist. Filing a patent application containing false technical information can lead to a lack of candor rejection or make the patent unenforceable in court.
  
• Inventorship & "Significant Contribution": Under current law, an AI cannot be an inventor. If the AI does too much of the "thinking" and the human inventor doesn't provide a significant contribution to the conception, the USPTO may determine the invention is not patentable at all.
  
• The "New Matter" Prohibition: Once you file an application, the USPTO generally forbids you from adding "new matter." If your AI-generated draft misses a crucial technical detail or includes an error, you may not be able to fix it without filing a completely new (and more expensive) application.
  
• Procedural Non-Compliance: Filing a patent involves strict formatting and "formal drawing" requirements. AI-generated images often fail to meet the USPTO’s specific line-art standards, leading to administrative delays or the need for costly professional revisions after the fact.

AVOID THESE TRAPS BY HIRING AN ATTORNEY WITH 18+ YEARS EXPERIENCE

The Do It Yourself Patent service is specifically designed to navigate these obstacles. By having a professional with 18+ years of USPTO experience perform a line-by-line review, you ensure that the AI's "rough draft" is transformed into a legally sound, defensible application.  

 Our opinion is that the following could arguably be the best practices for a DIY inventor:

• Focus on the "Human Element": Clearly document your own specific ideas and solutions. AI is best used to expand on your core technical concepts, not to invent them for you.
• Use AI for "Technical Padding": Use the AI to help write the general background of the field and provide alternative wording for technical descriptions to ensure the specification is robust.
• Perform a "Sanity Search": Ask the AI to help you find similar existing products or patents (prior art). This helps you understand if your idea is truly novel before you spend money on filing.
• Use private AI instances if possible: If you have access to secure, non-training AI models, use them to protect your trade secrets from entering the public domain.
• Fact-check every technical claim: Treat AI output like a first draft from a junior assistant. Verify every measurement, formula, and technical assertion it makes.
• Avoid inputting any "Secret Sauce" into public AI without knowing the risks: It is advisable to avoid putting the most unique, proprietary part of your invention into a public tool like ChatGPT or Claude because this can be legally viewed as a public disclosure, which may disqualify you from a patent.  So, do not do this unless you are comfortable with (a) accepting that the secret sauce will probably become public one day due to your disclosure,(b) following through with filing patent applications within the U.S.' one year grace period for public disclosures of your invention, and (c) you may automatically lose rights in foreign countries who rely on an "absolute novelty" patentability standard. 
• Avoid AI writing your final ownership claims: Claims are the most legally sensitive part of a patent. AI often writes them too broadly or too narrowly. Always have these professionally rewritten.
• DON’T list the AI as an inventor: Under current USPTO guidelines, only humans can be inventors. Adding an AI to the "inventor" block will lead to an immediate rejection.
• Understand AI might not be trained on the latest law: AI models have "knowledge cutoffs" and may not be aware of recent Supreme Court or Federal Circuit rulings that change how certain inventions (like software) must be described.

We believe that AI "novelty searches" are limited by their training data and often miss "non-obviousness" rejections—a common reason patents are denied.  While people can represent themselves, an attorney with 18+ years of experience understands the nuances of how USPTO examiners think, which cannot be said about a layperson representing themselves and which a language model cannot replicate. 

Practically, our opinion is  that "yes" it is safe to use AI tools if you plan to file with the USPTO and assuming you move forward with filing a patent application within the one-year grace period the US patent laws provide to inventors after any initial public disclosure).  Certainly, using public AI tools can be risky because your prompts may be used to train future models, potentially creating a "public disclosure" that ruins your patent rights in some countries internationally and in the U.S. if an application is not timely filed within the U.S.' grace period. We recommend as a best practice of using private, secure AI instances or drafting the core "secret sauce" manually before using AI for the general technical description. 

In our experience, AI speeds up the drafting phase, but it does not change the USPTO's examination timeline, which typically takes 18–24 months. However, a higher-quality, attorney-reviewed draft can reduce the number of "Office Actions" (rejections), potentially shaving months off the total time to grant. It should be noted that, if a fast patent process is desired, any inventor can pay a "Track one" fee for prioritized examination of their patent applications (which typically results in the USPTO examining the patent application within 3 months of filing).

Not at the moment. Currently, the USPTO and U.S. courts require that an inventor be a natural person. While you can use AI as a tool to help develop your idea, the legal rights must be attributed to a human who provided "significant contribution" to the invention’s conception. 

 The USPTO allows the use of AI for drafting, but it holds the person signing the documents—usually the attorney or the DIY filer—responsible for the accuracy of the content. They recently issued guidance emphasizing that human oversight is mandatory to ensure the AI hasn't "hallucinated" technical details or prior art. 

 Under current duty of candor rules, you must disclose any information material to patentability. While you don't necessarily need to state "AI wrote this sentence," you must ensure that the human inventor’s contribution is clearly defined so the USPTO can verify that a human, not just a machine, conceived the invention.  Practically, one can only assume that the USPTO expects AI is now or will eventually be involved in every patent filing submitted to it.

 We believe the most significant hurdle is "claim scope." AI often generates claims that are either too broad (leading to immediate rejection) or too narrow (making it easy for competitors to steal the idea). A professional review ensures your claims provide actual, enforceable legal protection. 

 No. If a human did not contribute significantly to the conception of the invention, it is not patentable under current U.S. law. Our service helps you identify the "human element" of your innovation to ensure it meets the legal requirements for a patent grant. 

 Technical errors in a filed application can be fatal to your patent. Once an application is filed, you generally cannot add "new matter," meaning an AI-generated mistake could permanently limit your protection or lead to a total rejection that cannot be fixed. This is one reason professional attorney review is considered by many to be a "best practice."

Yes.  And, understanding these terms is arguably crucial for anyone using AI to help draft a patent, as the USPTO uses them to determine whether your invention is actually yours—or just the machine's.

• Significant Contribution: This is the legal "gold standard" for AI-assisted inventions. Since an AI cannot be an inventor, you must prove that a human provided the creative spark or "significant contribution" to the invention's conception.
• Prompt Engineering vs. Conception: In the patent world, simply typing a generic prompt into an AI is often viewed as "problem identification," not "conception." Conception requires the human to have a specific, settled idea of how the invention works.
• AI Hallucinations: In a patent draft, a "hallucination" is when an AI creates technical data, citations, or prior art that doesn't actually exist. Filing these can lead to a rejection or even sanctions for a "lack of candor."
• Training Data Bias: If your patent relies on a specific algorithm, the USPTO may look at whether the "training data" used to teach the AI is part of the invention itself or just a tool used to achieve the result.
• Human-in-the-loop (HITL): This refers to the requirement that a human must be involved in the drafting and filing process. The Do It Yourself Patent service acts as your "professional loop" to ensure the AI's output meets legal standards.
• Model Weights and Parameters: If your invention is a new type of AI, you'll often hear these terms. They describe the internal "settings" of the AI. Whether these specific numbers can be patented is a complex, evolving area of law.
• Black Box Problem: This is the challenge of explaining how an AI reached a certain result. For a patent to be granted, you must be able to describe the invention clearly enough that someone else could replicate it—a requirement known as "enablement."