Summary
Our goal was to help you evaluate whether extending your declarative agent with an API plugin action is suitable for your scenario. We developed some criteria to guide your decision:
- The agent needs access to data outside its foundational model.
- The external data is exposed through an API.
- The API is implementing a supported authentication mechanism.
- The agent needs to read data in real time.
- The agent needs to modify the data.
- External data that the agent needs to access is structured.
We applied these criteria to the fictional scenario of a car repair shop that needs an AI-powered assistant to query information about repairs stored in an external system. Extending a declarative agent with an API plugin meets all requirements and leaves room for future requirements.
Using the criteria helps you evaluate whether using a declarative agent with an API plugin is suitable for your scenario. The criteria let you assess what kind of information the agent needs to access, how the information is exposed and how the agent needs to interact with it. Using these criteria gives you enough information to choose the right approach for your scenario.