Quick Facts
- Attorney fees for a non-provisional patent application average $9,086, pushing founders toward DIY filings that cost as little as $1,075 in government fees.
- A flawed patent disclosure cannot be repaired retroactively, meaning mistakes made at filing can permanently limit a startup's legal protection.
- Landfall IP, an AI-native patent law firm founded by Michael Drapkin and Nathan Mutter, launched in December 2025 with a Discovery Agent tool designed to structure inventor ideas before drafting begins.
Startup founders are turning to AI tools to file patents without attorneys, cutting costs in the short term while creating legal vulnerabilities that may not surface until a funding round or acquisition. The core risk is not the technology itself but what AI consistently misses: claim scope, disclosure completeness, and the strategic decisions that determine whether a patent can actually stop a competitor.
Michael Drapkin and Nathan Mutter, co-founders of Landfall IP, describe AI's patent capabilities as "jagged." The tools handle certain drafting tasks well but miss legal and commercial judgments that define a patent's real-world value. Both attorneys have represented Qualcomm, Micron, Coinbase, Oura, Salesforce, and Rubrik. Drapkin was named U.S. Patent Practitioner of the Year by Managing IP in 2022.
The Disclosure Problem
A DIY application may leave out technical details, design alternatives, and functional variations that become essential to future claims. Patent law does not allow founders to add that material after the fact. A founder who files early to lock in a date may later discover the document never provided enough support for the protection the company actually needs.
Claim scope is where the stakes are highest. Claims that are too narrow let competitors make minor changes and avoid infringement entirely. Claims that are too broad may overlap with prior art and get rejected or invalidated. A professional strategy targets what competitors must do to compete, then layers in narrower defensive claims as backup.
Hidden Risks Beyond the Application
Errors in assignment agreements create a separate category of exposure. Founders need to assign relevant work to the company and confirm they have no outstanding obligations to prior employers or universities. These gaps often go unnoticed until a due diligence review forces them into the open, at which point fixing them is expensive and sometimes impossible.
Drapkin and Mutter recommend securing assignment agreements when team members join, not when a deal is already in progress.
The Market Driving the Trend
The push toward DIY filing is real and financially understandable. Non-provisional patent applications cost between $8,000 and $18,000 in attorney fees at most firms. The DIY route runs $1,075 to $1,500 in government fees only. For pre-seed teams watching every dollar, the gap is hard to ignore.
Patent activity in AI is accelerating. According to WIPO, published generative AI patent families rose from 18,862 in 2024 to 37,808 in 2025, with more than 56,000 published over those two years. That volume exceeds the total from the preceding decade. Generative AI's share of all AI patenting climbed from 6.1% in 2023 to 8.7% in 2024.
What Landfall IP Is Selling
Landfall IP launched its Discovery Agent alongside the firm itself. The tool interviews inventors, structures their ideas, surfaces core inventive concepts, and generates a patentability report in real time. It is available free on a limited basis and is designed to help founders get clarity before spending money on drafting or filing.
The firm's broader platform handles drafting, quality analysis, consistency checks, claim-to-specification alignment, prior art assessment, and internal quality tracking. Drapkin previously led a patent prosecution practice at an Am Law 100 firm and co-founded Paximal, a startup focused on AI-assisted patent generation.
Both authors have a direct financial interest in the subject: Landfall IP competes for the same founders they warn about DIY risks. The underlying advice still carries weight. The most expensive patent mistake, as they put it, is discovering too late that a DIY application never adequately protected the innovation on which the business was built.
This article was written by an AI agent. Spotted an error? Send a correction and we will fix it.
