Our assistant just got way more helpful
A better assistant interface, custom requests, and more
This week we are proud to announce one of the biggest weeks of product in Edge’s history: a significant number of improvements for the assistant. This is one small step for Edge, one giant leap for patent-kind.
We are going to go over a number of new pre-made assistant features, leaps in usability, and the ability to ask Edge anything.
Our assistant is now way more powerful. But that’s not all. It’s also way easier to use.
Edge’s assistant now highlights text as it is being generated. You can then decide whether you want to accept, reject, or regenerate it instead of automatically inserting it in the body of the patent. If you want the regeneration to occur with specific instructions, you can do that as well. (If you have a version you like and want to edit it, you should accept it and then edit it further.)
On top of that, if you choose to try another version, you can easily flip between multiple options to compare before you select which one go to with.
You’ll also notice that the assistant automatically inserts references. It won’t just say “Claim 1” but actually insert an auto-updating reference with our “@” tag feature. That way, if you make a change like deleting a claim, the assistant-generated text will remain consistent.
New writing powers
Edge’s assistant was previously empowered to help you draft your claims.
Edge is now able to create several major types of content. The assistant can now write the summary, abstract, field, background, and claims.
You can also give specific instructions when using one of our commands. Normally, you would just click on a command after pressing command-k. Instead, type your additional instructions in the text box, click the command you want to use, and let it rip.
Hint: next week’s new commands will relate to the detailed description.
But what if you want to edit your work, or work with the assistant free-form? Read on.
Edge now supports free-form writing. Unlike with our built-in commands, the assistant does not know what to do by default; you have to tell the assistant what you want. What it does do is write like a patent attorney, so you don’t need to “prompt engineer.” Just tell it what you want to do.
You can call up the assistant and ask it to generate arbitrary text. For example, suppose you want to create a boilerplate embodiment paragraph but know that it may be lengthy. You can use the assistant to save you time and draft the paragraph itself.
You can also make changes based on specific text. Just highlight text to ask the assistant to edit text or add on to text.
Editing text modifies the existing text. You need to tell the assistant specifically what you want to change and it will do so. This is a good use case, for example, if you see a short figure description you want to change.
Adding on to text adds additional text to the patent with the highlighted text as additional context. So for example, if you want to add multiple dependent claims to an independent claim, highlight the independent claim and tell it what you want it to do.