Quick Review

In the previous course, we introduced Teams, Agents, and Tools. A team represents a group of agents working together toward a shared goal, with each agent responsible for a specific step in the workflow. Within this structure, agents coordinate decision-making, while tools provide the concrete capabilities agents rely on to take action. Both agents and tools can be system-provided or custom-built, and together they enable flexible, extensible AI-driven workflows.

In this course, we focus on two system tools: GetNamedUserItemsTool and GetRelatedItemsTool. The GetNamedUserItemsTool identifies a named user item by interpreting natural language input and resolving it to a specific item defined within the project. The GetRelatedItemsTool retrieves related items from a collection by constructing and executing a query based on the user’s input. When used together, these tools enable an agent to first resolve the correct named user item and then retrieve the associated data from the relevant collection.

These system tools are commonly used within a team, where agents share context and pass results between one another through background instructions. By assigning clear responsibilities, such as intent interpretation, item resolution, and data retrieval, to individual agents, teams provide a structured and scalable way to combine agents and tools into coherent, data-driven interactions.

Finally, we will take a closer look at knowledgebases. A knowledgebase is a vectorised file containing a collection of useful information that a language model can search to answer questions more accurately and with greater context.

---