A Kick Start to Conducting Invalidity Searches

NLPatent's Patent Number Search offers an efficient way for practitioners to conduct invalidity searches using NLPatent's proprietary Large Language Model (LLM).

An overview of conducting invalidity searches in NLPatent is provided in the video guide + by text below:

1) Start a New Search

Log in to NLPatent.
Click New Search.
Change the default search title (e.g., to Invalidity or Whatever you'd like!) and click the ✅ checkmark.
Click Patent Number and add the patent you'd like to challenge
Since we're working on an invalidity case, tick the "Filter out patents filed after this patent's priority date" box.

Tip: We generally recommend using the Patent Number search option as the AI looks at the claims in the context of the specification. That said, some users do like to use Natural Language and then paste in the claim they're trying to invalidate.

2) Result Organization

Once you click start a search, the results will return to you with the top 100 most relevant results, ranked from most to least similar by default. You can configure how many results you want to see from the search set up page.

3) Filter Results (As Necessary)

By default, the search covers all jurisdictions and patents in the database (excluding the ones published after the priority date of the patent you have selected). If you'd like to selectively look for specific time frame, etc. you can filter out the results.

4) Use the Ask NLPatent Function

Historically, document review takes a sizeable amount of time. This is where you can use Ask NLPatent to ask specific questions across your result set, reducing the amount of time spent on reviewing documents. In particular, a query we like using in the case of invalidity is:

Generate a table comparing this patent to (Patent # we want to invalidate) US11794018B2. Could this patent be used to invalidate (Patent # we want to invalidate) ?

Once you've submitted your question, the answer will pop up below the result like so:

To learn more about prompting in Ask NLPatent read this article

5. Relevance Analysis (Optional)

You can also optionally use Relevance Analysis to generate a summary on why your result is similar to your search query.

6) Refinement Search with Highlighted Text

We always recommend doing 1-2 refinements per search to ensure that you are not missing out on anything important. To do a refinement:

Highlight any relevant text within a patent and click Save this section.
Then click Refine to update results based on the highlighted content.
Highlighted text appears in yellow, and the Refine button will flash when ready.

After refinement, arrows appear beside each result showing which patents moved up, down, or are newly relevant.

Tip: You can also do a refinement of a search with Non-Patent Literature and other patents you might not see in your search results. See more on Refinement Here

7) Optional: Use Keyword Filters and Highlights

Optionally, you can use Keyword Filters to include or exclude specific words (e.g., cardiac).

To visually highlight terms without changing results, use Edit Highlight.

Example: Enter implant* to highlight implant, implanted, implantable, etc.

Keyword Filter changes the results shown.

Highlight acts as a visual “find” tool within documents.

To read more on Keyword Filtering and Highlighting check this article.

8) Export Results

Finally, users can download the results by going to the download button in the top right corner. Select the appropriate fields you'd like to export and then click Generate.

If Relevance Analysis is selected as a field, only patents where a relevance analysis has been generated will contain the text in the report.

Extra Tips

Enforcing a Claim:

If your patent has several claims and you want to re-enforce a specific claim(s), you can manually add the patent you're trying to Invalidate. Then, highlight the claim(s) you'd like to use and hit refine, this way a specific claim is emphasized over the others ones, which can be helpful to drill down specific concepts.

Query Combinations:

Users may also find it helpful to use the Multiple Queries and Query Combinations to function to run several searches in tandem (ex. Patent Number, Claim 1, etc.)