How Amplified learns from your search results

We do not use your data to train our foundation models. However, Amplified can use the relevant patents that you find in a project to go and find more like them. To learn in a project, make sure that Learning from Relevant results is on.

There are two ways to add relevant results to a project. You can mark a search result as Relevant or you can upload a patent directly by number.

Saving relevant results

To save a result as Relevant click on the green checkmark next to the result.

After you've saved a result as relevant the Update results button will turn red. When you're ready, click on Update results and any new relevant results will automatically be used to re-sort the result list.

How to undo learning and return to a previous result list

Amplified only uses the results currently marked Relevant. So you can always go back to a previous result list by changing a Relevant result to Unreviewed or Flagged and then clicking Update results.

You can also temporarily turn learning off entirely by clicking on "Relevant patents" under Learning from and changing it to Off. When learning is off, Amplified will only use your project inputs to search and sort.

Uploading relevant patents by number

You can use the Upload button in the header to manually add patents to your project results. You can add patents as Relevant, Flagged, Hidden, or Unreviewed. All Uploaded patents are saved in your Uploaded list. Relevant, Flagged, and Hidden patents will also be in your Reviewed list.