Scite MCP Patents Demo

hi there.

I'm Drew Barantini, VP of Product at Scite, and I want to walk you through how you can search patents inside of our MCP.

Now, we launched the Scite MCP with access to our full range of academic literature, and now we're expanding that to include additional data sets like patents.

So inside of any tool that you're using, like Claude or ChatGPT or other tools that connect to to MCP servers, you can ask a plain language question.

And I'm going to search this here and run it.

So I'm searching for patents on MRNA lipid nanoparticle delivery systems sorted by forward citation count.

Now, I'm asking it specifically to present the top five patent families in a structured table with specific data.

So this is going to load the tools, it's going to use our Scite MCP, it's going to search those patents, and then it's going to give us this table that prints out all the information as I requested.

So it's showing the top five patents here related to this information with key assignees, the filing date and forward citation count.

It obviously provides additional context here and it talks about who holds the most influential IP around this specific area of study.

So you can see here showing that additional information and it's going to provide that.

So this is looking at all of that patents information we have, which is over 90 million patent families and 200 million patent applications across a broad aggregate data set.

And the benefit of using the MCP in a tool like this is you can also ask questions related to the academic literature so we can look at papers and patents and understand how they relate.

So I'm going to ask it to search the literature for papers on that specific delivery topic and sort it by supporting citations to understand what, what's the most validated research and understand if there's a correlation between any key assignees from the patents that appear as authors or any affiliated institutions in those top papers.

So this is going to leverage a different tool, the tool that we launched our MCP with to search that literature and it's going to find that information here, it's going to orchestrate the calls, it's going to get data back from Scite on all of that rich context that we have, all of our smart citations, over 1.6 billion of those, and then it's going to build that information and respond as we expected.

So you can see here, it's now showing the top, MRNA LNP delivery papers and it's ranking them by supporting smart citations.

So you can see number one here we're showing paper published, dates and we're seeing lead authors, we're seeing supporting cites and total cites.

So this is showing out of the 2,000 plus total citations for this individual paper, 44 of them, we have marked as supporting that information any of those sites.

And you can see here it's going to show you all that information and it even calls out the overlap here, which it says is striking.

So it's showing a specific information about three of the foundational papers coming directly from this group.

Scientific founder.

So it's giving you information to summarize that data.

Now you can keep asking questions here, follow the thread.

You can mix and match across papers and patents to bring the research together, making sure that you're enriching that data and your understanding of the research landscape with smart citations.

Now another really interesting thing here too is you can also take this further.

So on scitedemos.com we're sharing all the different outputs and artifacts that we're creating out of Claude specifically.

And these are really rich examples of how you can combine this data.

And we're including the actual prompts that were used to generate these with downloadable PDFs.

So for this example here you can see this HTML report that was generated based on our site smart citations of all the literature as well as our Scite patent information.

And it's tracing this path of this 2005 paper around MRNA and how that is being used across the landscape into the period of time with COVID and how this has been built on.

And it's giving all of that data backed by information from site, it's getting interactive charts, it's creating all of this data and it's building this.

And this happened in a matter of minutes.

So this is five to 10 minutes to create this really rich interactive artifact that pulls together all the research in a way that is visual and easy to comprehend and understand.

And ultimately the sky is the limit with what you want to do inside of tools like ChatGPT and Claude to generate rich artifacts from a wide range of, of literature and data and research like what we have at Cite with patents and papers.

As always, if you have any feedback on anything and how we can improve this, please let us know.

We're continuing to improve Scite as an entire platform, including our MCP offering, and would love to hear more about how you're using it today.

Thanks.