Generative AI represents the first foundational technology of our time with the unique ability to simultaneously enhance quality and reduce costs in high-value knowledge work. This dual capability enables business-of-expertise professionals, such as patent attorneys, to scale their output and save time, all while maintaining the excellence of their work.
But the transformation doesn’t end there. Generative AI catalyzes a deeper change—one that challenges the traditional division of labor between in-house patent departments and external patent firms. In this article, I argue that AI is empowering clients to take a more proactive role in the patent drafting process, reshaping the relationship between attorney and client, and—crucially—resulting in significant savings in external attorney hours.
My Thesis:
AI dissolves the traditional division of labor between in-house patent departments and external patent firms, saving attorney hours while maintaining the quality of the work.
Traditionally, external patent firms have played a dual role: they serve as both strategic advisors and the executors of the patent application process. For instance, when a new patent application is required, the external patent attorney advises on claim scope and strategy, while also drafting the application itself. The inventor’s role typically extends only to providing an invention description, which the attorney then uses to develop the patent application.
Generative AI, however, is beginning to disrupt this traditional model. With the right tools, clients can now take on substantial portions of the work themselves—drafting sections of the application that are remarkably close to the final product. This eliminates the need for multiple translation steps, from idea to invention description to patent application.
More importantly, so I argue, this shift represents a profound transformation in the patent drafting process. It is not merely an incremental improvement—it empowers patent applicants to take a much more active role, reducing time spent on back-and-forth communication and saving external attorney hours.
A Low-Hanging Fruit: Writing the “Background” Section
One of the most immediate and accessible ways to benefit from generative AI in patent drafting is through the creation of the “Background” section. While this may seem like a small part of the patent application, it plays a critical role in setting the stage for the invention.
The beauty of this shift is that inventors, equipped with the right AI tools, can now create a high-quality first draft of this section themselves, saving approximately an hour of patent attorney time in the process.
Let’s break this down.
Step-by-Step Guide to Writing the “Background” Section with AI
In the following guide, the inventor only needs to write down three key bullet points, and the rest is handled by the AI. My experiments show that the result is a “Background” section that is more than sufficient for many cases or, at the very least, provides a great first draft for further refinement and expansion:
1. Define the Domain
Clearly identify the relevant technological field or industry. This provides context and helps the reader understand where the invention fits within the broader landscape.
Example: “Domain: Robotics, specifically robotic arm control and teaching.”
2. Describe the Status Quo
Briefly summarize the current state of the art, including existing technologies and their limitations. This demonstrates awareness of the field and highlights the need for a new approach.
Example: “difficult to code, difficult to make the robot arm dexterous and adaptable”
3. State the Problem
Clearly articulate the problem your invention addresses on the most general level. This emphasizes the gap your invention fills and creates anticipation for your solution.
Example: “make the teaching of the robotic arm more intuitive”
4. Prime Your AI Chatbot
Pick an AI chatbot that aligns with your confidentiality requirements. Before asking the AI to generate the draft, ensure that it is set up to take on the role of an experienced patent attorney. Establish expectations clearly, as follows:
Prompt: "You are an experienced European patent attorney specializing in [your Domain bullet point]. Your task is to help me write an expertly crafted patent document for my invention. We will proceed step-by-step, with me providing instructions and you generating the corresponding sections of the patent document. Adhere closely to my guidelines and ensure your responses are in line with patent drafting best practices."
5. Write with AI
You can now prompt the AI to generate a draft of the “Background” section using the bullet points you’ve crafted. Here is the prompt to use:
Prompt: "Write a concise and informative “Background” section for the patent document based on the following information:
Field of the Invention: [Domain]
Limitations in the Status Quo: [Status Quo]
Objective of the Invention: [Big Problem]
Adhere to the following guidelines:
1. Start with a concise paragraph explaining the field of the invention.
2. Describe the limitations in the status quo in a few paragraphs, without revealing any details of the proposed invention. Keep the descriptions brief, to the point, and relatively general.
3. Avoid any statements that imply certain technical features were “known” from “the prior art.”
4. Conclude with a paragraph stating the objective of the invention using the phrase “It is therefore an objective of the present disclosure to...”. Formulate this objective in a general manner. When in doubt, use “...to overcome the above limitations at least in part.”
5. Write in continuous text without any sub-headings.
6. Maintain a formal, objective, and technical tone throughout the section."
The Result
This is the output of a local Llama 3 8B Instruct model in GPT4All following the above guide (learn how to set up a local AI in this free course):
The field of robotics has witnessed significant advancements in recent years, with robotic arms becoming increasingly prevalent in various industries such as manufacturing, healthcare, and logistics. However, despite these advances, the teaching and control of robotic arm movements remain a challenging task.
Currently, programming robotic arms requires extensive knowledge of computer languages, algorithms, and mechanical systems. This complexity can lead to difficulties in achieving precise movement control, resulting in reduced dexterity and adaptability of the robot arm. Furthermore, traditional methods for teaching robotic arms often rely on trial-and-error approaches or require significant expertise in robotics engineering.
The limitations inherent in current robotic arm control and teaching methodologies hinder their widespread adoption and utilization. The need for more intuitive and efficient ways to teach and control robotic arms has become increasingly apparent.
It is therefore an objective of the present disclosure to overcome these limitations at least in part, providing a novel approach to robotic arm control and teaching that enables users with varying levels of expertise to effectively utilize robotic arms in diverse applications.
In my opinion, this output generated by the AI using the above guide is more than sufficient for many patent applications and can be produced in mere seconds. While it may require refinement for particularly important cases, the AI-generated “Background” section provides a strong starting point, with clear and concise language that sets the context for the invention.
Download Your Free Resource
The guide above will be part of a more extensive workbook I’m currently developing to help non-attorneys craft key sections of a patent application using AI. I’m offering you an exclusive preview of the workbook’s chapter, “A Springboard for Your Invention,” which contains a detailed guide for drafting the “Background” section with AI.
To get your hands on the free preview chapter, simply subscribe to the Powerclaim Newsletter below, and I’ll send it directly to your inbox:
Never stop exploring!
Basti