We’re excited to announce the launch of AI Reports for Patents, our newest feature designed to uncover unique and valuable insights from the patent dataset. With this feature, you can build downloadable reports tailored to common patent-focused research objectives, such as a Commercialization Report, Citations Impact Analysis, Technology Clustering Report, White Space Anlaysis & more.
Additionally, you can create custom report templates (or custom prompts) by defining your objective and selecting the specific data points you’d like to analyze. Custom prompts can be saved to dashboards for easy access and future use.
We believe this feature will transform how users conduct research on Cypris. AI Reports allows users to run in-depth queries, analyze the accuracy of raw results, and use that data to create highly detailed, targeted reports. This process not only enhances users' confidence in the information provided by the AI model but also enables them to focus the AI on a defined pool of data.
This is just the beginning for AI Reports. In the future, we plan on extending the feature to our other datasets like Research Papers, Market News, and Organizations, but for now, we trust and hope you can find value for your patent-focused research initiatives.
Why We Built This Feature
One of our most popular product offerings over the last year has been our Research Briefs, created by Cypris analysts. Time is proving to be one of the more valuable commodities for R&D teams, and traditional dashboard products have led them to spend more time on research and less time on development. This is a misstep on the vendor's part and is slowing down critical R&D initiatives.
Research Briefs have been well-received because they give valuable time back to researchers. With AI Reports, we set out to bring that same value to our self-service dashboard.
We chose to start with patents due to the high volume of patent-related work our analysts have spent time on. We then analyzed the most common patent report types users were manually building to determine which ones to include as suggested prompts. Each suggested prompt reflects the optimal report structure, informed by our findings and close collaboration with users during testing.
Additionally, we wanted to provide agility in the feature by allowing users to create and save custom prompts, by simply adding an objective and descriptive formatting.
User Needs Addressed
- Automatically generate reports using the following suggested prompts:
- Patent Commercialization Reports
- Evaluates a patent's commercial potential, highlighting sectors or companies where the technology could be monetized through licensing.
- Citation Impact Analysis
- Analyze a patent's impact based on its forward (cited by) and backward (cites to) citations.
- Technology Clustering
- Examines where patents can be grouped according to applicable technology sectors.
- White Space Analysis
- A white space analysis that identifies gaps in the patent landscape, highlights innovation opportunities and provides actionable recommendations supported by data insights.
- Patent Portfolio Analysis & Impact Assessment
- A comprehensive analysis on the selected patents, their owners, and their possible impact on the market.
- Design custom report prompts by describing an objective and source formatting. Save prompts in your dashboard for quick future use.
- Suggest edits to the report using the ‘Regenerate’ text box.
- Download reports to PDF.
- Patent Commercialization Reports
How AI Reports Work
- Run any type of search (semantic, boolean, or organization).
- Navigate to a patent in the results
- If on an organization page go to their patents tab.
- If you’d like to build a report on 1 patent, select the AI Report Blue Icon to open the report builder.
- If you’d like to build a report on multiple patents, checkbox the group of patents and select AI Report in the bottom pop-up bar.
- Choose a suggested prompt or Design your own prompt.
- Once you get back your results, download to PDF or choose to regenerate the report with added comments.
How We Built This Feature
This feature is powered by ‘retrieval augmented generation’ or ‘RAG’.
Explained simply, RAG is an AI technique that focuses the answers you get from language models by combining external contextual knowledge with text generation. When you ask a question, a RAG model searches through a focused set of documents (or ‘context’) to find the most relevant information. It then uses this information to generate a response, effectively blending up-to-date data with its language skills. This means you get more accurate and context-aware answers, making the AI's responses more accurate and reliable.
Cypris RAG Context for This Feature
For this feature, we’re primarily looking at the selected group of patent documents as the initial context to build the analysis, and depending on the report type retrieve answers from the Cypris database in addition to external providers.
Depending on the chosen report type we’re analyzing a different breadth of context from the patent document. For example, the citations impact analysis primarily looks at the patent’s forward and backward citation data.
In ‘Design Your Own Prompt’ the user can specify the context to be either the Metadata or Full Text.
For AI Reports case, Metadata includes: Title, Abstract, Assignee, Country, Publication Date, and Inventor. And Full Text includes Claims and Citations + all Metadata.
Finally, certain prompts are restricted by a maximum amount of ‘Input Context’. For example, when doing a Citation Impact Analysis, it can only be done on a maximum of 20 patents due to the breadth of input context.