How can AI be used to systematically develop patent portfolios, analyze patent literature, perform feature comparisons and support the entire invention process? These were the questions discussed in the final module of the MIPLM (Master for Intellectual Property Law and Management) Master’s program. The associated case study was presented as part of the IP Management training for the joint Master Program of CEIPI & the European Patent Office. Various AI tools were used during the lecture, in particular for AI-supported patent landscaping patentbutler.AI from uptoIP & ABP Patent Network was used. You can find more information on the chosen scenario and application in the case study at I3PM.

What is meant by synthetic invention?

Synthetic inventing is the basis for specifically created spheres of exclusivity as part of the development of a 360° IP strategy, where the IP needs are systematically identified and characterized. For this purpose, fields of action are defined in which the creation of strong exclusivity positions through IP leads to a high strategic benefit by applying the IPFD methodology and a fundamental understanding of the target business model. Also, goals must be formulated which should be realized using IP in the business model. This shows which types of IP rights are needed and what they refer to. However, especially for patents, the specific inventions which should be protected with IP are only in rare cases obvious at this point.

The implementation of a 360° IP strategy regularly requires the competence to systematically identify inventions in the desired places by a white spot analysis and to draft patent applications based on these inventions. This procedure can be described as synthetic, as it does not start with a specific invention, as in the classic patenting process, but rather focuses on the need for prohibition, which is realized by patenting specifically created inventions.

What is AI-aided synthetic inventing?

Artificial Intelligence (AI) has become an indispensable tool in the realm of synthetic invention, revolutionizing the way we identify new solutions to specific technical problems in a targeted way. Through advanced algorithms and machine learning techniques, AI accelerates the process of designing and optimizing synthetic inventions, leading to a fast identification of a broad range of options for innovation into technology fields, which are still unoccupied by competitors and patenting positions.

Practically, this can be done by letting an AI generate new inventive solutions based on known innovation principles, e.g. with TRIZ. TRIZ is a systematic methodology for problem-solving and innovation developed in the former Soviet Union by Genrich Altshuller and his colleagues. TRIZ is based on the idea that there are universal principles of innovation that can be applied across different fields and industries.

In the case of the smart parking use case, a possible starting prompt for an AI in AI-based systemic inventing is: “Act as an experienced engineer and inventor familiar with the TRIZ principles. Apply the TRIZ principles to find new, and for an expert in the technical field non-obvious solutions to the following problem: A car with a human driver is supposed to park in a parking garage at a certain spot. This spot is not permanently marked, it can be a random spot in the parking garage. Focus on the following detail: What inventive ways are there to directly inform the driver or the car where the spot is where the car should park? Please describe the necessary system components and how they work together to solve the problem technically, such as lights integrated into the floor to inform the driver. Do not describe a solution that already appears in full in the prior art and patent literature. Describe 10 inventive solutions to the problem. For each solution, state which TRIZ principle was used in the solution.”

The answers will be different between various AIs, but here is an illustrative answer:

Solution 1: Dynamically Illuminated Parking Spots

  • TRIZ Principle: Segmentation (Dividing an object into independent parts)
  • System Components: LED tiles embedded in the parking garage floor, a central control system, and in-car display connected to the garage system.
  • Working: The central system assigns a spot and illuminates the LED tiles forming an arrow pointing to the assigned spot. The in-car display confirms the spot and mirrors the arrow directions.

Solution 2: Augmented Reality Navigation

  • TRIZ Principle: Taking Out (Removing a troublesome or harmful part of a system)
  • System Components: Smartphone app with AR capabilities, parking garage cameras, and a cloud-based coordination system.
  • Working: The app interfaces with the garage system, which assigns a spot. Using the garage’s cameras, the app overlays a virtual path on the real-world view through the phone, guiding the driver to the spot.

…

Those new solutions have so far only applied a single TRIZ principle at a time, but also multiple principles may be used at the same time. When appending the prompt with the following line, “For each solution, use more than only one TRIZ principle and state which TRIZ principles were used in the solution.” you will receive more complex new solutions such as:

Solution 1: Gamified Parking Experience

  • TRIZ Principles:
    • Transformation of Properties: Turns parking into a game.
    • Humor: Incorporates playful elements to make parking enjoyable.
    • Psychological Inertia: Leverages the human tendency to follow game rules and complete tasks.
  • Components: In-car display with gamification elements, sensors in the garage, and a central control system.
  • Working: Parking is presented as a game with points or rewards for accurate parking, time taken, etc.

Solution 2: Drone-Assisted Parking

  • TRIZ Principles:
    • Mediator: A drone acts as an intermediary between the car and the parking spot.
    • Copy: The drone replicates the desired parking position for the car to follow.
    • Pneumatics and Hydraulics: The drone could potentially use air currents to nudge the car into the correct position.
  • Components: Parking assistance drones, a central control system, and communication between the drone and the car.
  • Working: A drone is assigned to a car and flies to the designated spot. The car follows the drone’s movements to park accurately.

AI-assisted analysis of the patent landscape with patentbutler.AI by uptoIP® & ABP Patent Network

An AI-assisted white spot analysis of the patent landscape can be setup within the patentbutler.AI tool with the following few steps. In the first step the initial features of the desired patent, which should lie in a white spot need to be defined within the system as a starting point. This can be done by the “create features” function of the tool. An initial set of features for the use case of “customized digital parking” can comprise e.g.:

  • Sensors in each parking slot detect vehicle presence, providing up-to-the-minute available data
  • LED displays direct drivers to the nearest available parking spaces
  • Users can view real-time availability, reserve spots, and receive navigation assistance
  • Cameras capture and log vehicle plates for seamless access control and billing
  • Adjust parking fees based on demand, time of day, and occupancy rates
  • Motion sensor-controlled lighting adjusts brightness based on occupancy to enhance safety and save energy

Once the features are defined, any set of them can be used to identify patents with similar features within a chosen patent portfolio, in this case a portfolio relevant for digital parking solutions. The found patents can then be analyzed based on the similarity and relevance of individual features or combinations of features. The similarity is indicated by the shade of the green background color for each feature, with dark green meaning a high similarity and white less than 50 % similarity. To get more information about the similarities found in the patent document and to get a more concrete understanding of the results the patentbutler.AI tool provides detailed information for any compared feature, e.g. where the information can be found in the original patent documents.

Based on the found similarities of the initial set of features with the patent literature, in the next step features must be edited, deleted, added or combined to feature combinations to identify features (or feature combinations), which are less covered in existing patents. For example the above chosen set of features can be modified by regrouping the features to multiple feature combinations and addition of new features:

  • Feature Group 1:
    • Sensors in each parking slot detect vehicle presence, providing up-to-the-minute available data
    • LED displays direct drivers to the nearest available parking spaces
    • Users can view real-time availability, reserve spots, and receive navigation assistance
    • Cameras capture and log vehicle plates for seamless access control and billing
  • Feature Group 2:
    • Adjust parking fees based on demand, time of day, and occupancy rates
    • Motion sensor-controlled lighting adjusts brightness based on occupancy to enhance safety and save energy
  • Smartphone assisted

This process of regrouping, editing, adding and deleting features can be iterated until a desired degree of low similarity compared to the existing patent literature is reached and a white spot is identified.

Here is a short introduction video how to use AI prompting for IP-design:

About the lecturer:

The student exercise was led by Dr. Hannes Burger. Dr. Hannes Burger is a long standing supporter of the MIPLM education at the CEIPI and acted as a co-examiner in the MIPLM since 2022. Dr. Burger has been specialized in intellectual property rights for many years and is nationally and internationally active in prosecution as well as litigation regarding the field of IP. Hannes studied law at the University of Linz and is Attorney at Law in Austria, admitted to the bar in Germany, as well as European Patent Attorney. This double qualification is quite rare in the Europe and especially in the German speaking countries, where only a handful of Attorneys at Law hold the qualification before the European Patent Office. Hannes has almost twenty years of experience in cross national IP litigation and enforcement, and advised clients in a wide range of industries, from mechanical engineering to medical devices, chemicals to biotech.