Skills

This video provides a walkthrough of skills in Patsnap Eureka's Hero AI assistant, explaining how a skill acts as an instruction set that defines what data to search, how to analyze it, and how to present the results consistently. You'll learn the three ways to access a skill—typing a forward slash, using the plus icon to browse the library, or simply asking a question and letting Hero auto-match it—as well as how to inspect a skill's details before using it. The video also walks through a real example using the competitive landscape skill to generate a structured report, and shows how to export or share results, and how to build your own custom skill when your team's process is stable. By the end of the video, you'll understand how skills turn one-off AI answers into consistent, repeatable workflows for R&D, IP, and competitive intelligence teams.

• Treat a skill as an instruction model that guides the AI on how to respond.
  
  A skill defines three core elements:
  
  What to search: the data sources to query, such as patents, scientific papers, clinical trials, or company data.
  
  What to do with the data: how to analyze, compare, and rank the information.
  
  How to present the output: the report structure and format so the result is actionable.
  
  Use skills when you need consistent answers instead of one-off prompt-dependent results.
• Recognize that skills reduce inconsistency in AI outputs.
  
  Use a skill when multiple team members ask similar questions and need the same method and rigor every time.
  
  Apply this approach in functions such as:
  
  R&D
  
  IP
  
  Competitive Intelligence
  
  Any team that repeatedly asks similar research questions
• Open the skill library in one of three ways:
  
  Type / in the input box and search the skill library directly.
  
  Click the plus (+) icon to browse available skills.
  
  Ask your question normally and let the system automatically match it to the most appropriate skill.
  
  Choose the method that best fits your level of certainty about the skill you need.
• If you are unsure what a skill does, click the customize icon on the left side.
  
  Review the skill breakdown to understand:
  
  What the skill is designed to do
  
  When to use it
  
  What information was used to create it
  
  How the output is intended to be used
  
  Confirm the skill matches your use case before proceeding.
• Identify the business question or research need you want to solve.
  
  Example use case: competitive analysis.
  
  Click the plus (+) icon and go to New Skills.
  
  Select the relevant skill, such as Competitive Landscape.
  
  Enter your use case or technology area of interest.
  
  Submit the request to activate the skill and generate the analysis.
• Check that the output is structured and aligned to the selected skill.
  
  Look for key sections such as:
  
  Executive summary
  
  Competitive arena overview
  
  Visuals/charts
  
  Evidence boundary
  
  Confirm the report is easy to interpret and supports decision-making rather than presenting raw text only.
• Use the available export options to distribute the output:
  
  Copy Markdown
  
  Export PDF
  
  Export Word
  
  Share the report through a link if you need to collaborate with:
  
  Colleagues
  
  Third-party services
  
  Use sharing to support team review, collaboration, and downstream decision-making.
• If your team has a unique internal process, create a custom skill instead of relying only on the repository.
  
  Click the option to Create with Eureka.
  
  Build the skill by providing detailed instructions, including:
  
  Non-negotiable requirements
  
  Desired inputs
  
  Preferred analysis method
  
  Desired output format
  
  Be as descriptive as possible so the skill reflects your internal workflow accurately.
• Use skills to define what to search, how to analyze, and how to present results.
  
  Start with the / search or automatic matching when exploring a new question.
  
  Build a custom skill once your process becomes stable and repeatable.
  
  This approach helps convert one-off questions into repeatable intelligence workflows.