How Robotic Patent Drafting (RPD) Helped Me Reduce Patent Delivery Time From Weeks Or Months To Days

My journey as a patent attorney began at a time when software patents were just emerging. That era laid the foundation for what is now a significant shift in patent drafting – the advent of Robotic Patent Drafting (RPD).

How RPD works – important: the Input Data

In short, a patent drafting bot works like any other piece of software: the user provides input data and information about the desired output data. Once this data is provided, the software generates the output data, which can and should be reviewed and modified by a human. In the case of RPD, the input data often includes not only information about the invention, but also data about the structure of that information, such as which figures show which elements of the invention. My very first patent drafting robot generated pictorial descriptions of gear variations stored in a spreadsheet file, saving me a lot of time in writing an enabling patent specification.

Some RPDs go beyond this traditional way of working and would provide additional features such as automatic detection of substructures and sections of devices, often from interpreting patent claims after they have been entered into an online form or text document intended to be sent to a server. This can be used to generate figures illustrating claims automatically.

Of course, all modern RPDs have features that highlight quality problems in the input data and issue warnings if common error patterns are found. Building a set of patent claims in a good online tool can be an instructive lesson for beginners and a therapeutically refreshing experience for experienced patent drafters.

To be more specific, depending on the required quality of the resulting patent application, I spend about 45 minutes (for an economy class patent application) to 8 hours (for a litigation grade patent application) generating the input data for each robotic patent drafting tool.

Check Carefully: The Output Data

To cut a long story short, each RPD bot generates an output text file that needs to be carefully reviewed, because auto-generated texts can contain incredible nonsense wrapped in flawless language, similar to what MBA students, management consultants, and politicians often produce. In my case, it takes about two to three times as much effort to correct and modify the output text file as it does to generate the input data. What I often do is to check the output text file for language inconsistencies and logical errors, and then re-run the RPD generation after making changes. The total effort is still only 20% to 30% of the time I used to invest in writing patent applications the conventional way, and with much higher quality.

How to get started with RPD today

There are two ways to start using patent drafting automation tools:

a) read articles, books, and watch videos until you feel comfortable with your needs, followed by trying a few patent drafting bots yourself, or

b) hire a professional trainer (me) to help you shorten the learning curve (click here).

I would recommend option b) because that trainer will also help you choose the right patent drafting robot. As I write this article, there are seventeen (17) commercially available bots, and a new one comes out every month or so.

Continue reading here.