Edge has some Assistant magic for figures
As every patent practitioner knows, figures are a key part of a patent—and a major part of the work. In a previous update, we showed how we are taking the pain out of label generation and referencing. And last week’s update focused on how we have been making our assistant way more powerful.
Now, we are proud to announce a major update that brings those two elements together.
We are introducing two new features: AI-image generation and AI-assisted figure descriptions. Let’s dive in.
The world has been captivated by DALL-E and Midjourney. After all, patent draftspeople can cost a fortune at scale and delay a patent application by weeks, affecting your priority date.
A number of our users have reached out wondering when they would get the same functionality in Edge.
That day is today.
We’re proud to announce our text-to-image generator. It works very simply:
In the figures tab, instead of pressing “upload,” press “generate” instead
Type what you want to see in a textbox. More descriptive is better. Note that we have designed it to do what is needed, so don’t try to prompt engineer. Just describe plainly what you want the image to do
Wait a bit and in about a minute (sometimes less), you will get four images to choose from that are USPTO submission-ready. We will automatically color-adjust to make sure it is perfectly black-and-white
This image generator is best used for conceptual or demonstrative figures. For now, the Generate tool will not generate formulaic figures, like flowcharts or chemicals with Markush groups. But don’t worry, we’re already looking into that!
As for CAD drawings, you cannot yet upload them and get a line drawing out. However, note that most CAD drawings will do that for you. For those types of drawings, we recommend using the export line drawing functionality most CAD software supports and uploading it to Edge to start annotating with our labeling tool.
Detailed description for figures
Edge’s Assistant is getting more capable every week. This week, we shipped the first of our Assistant functionality for figures. We have much more in the pipeline, but we saved the best for first: you can generate detailed descriptions of drawings.
Our written description tool allows you to select a figure and ask the Assistant to generate a full written description. It will use every label in the proper format. (Soon, it will also create them with “@” tags, just like with claims.) The Assistant also has the context of the disclosure and the claims to know what the technology does as well as what it has to enable.
Here is how to access this command:
Go to the Detailed Description section
Press command-k to invoke the assistant
Select “Generate written description of a figure”, and add any additional instructions if you have them
Select the figure you want to describe
The Assistant will generate a full written description for you
That’s it! There’s much more coming down the pike for the Assistant when it comes to figures. We know it is one of the most painful, time-consuming parts of enabling claims.