Open topic with navigation
Agents automatically find documents for you that match your interests. A user who is interested in football and gardening could, for example, create a Real Madrid agent and a Pest Control agent.
When you create an agent, you give it training text. This training provides an example of the type of text that the agent must look for, so that an agent returns only documents, profiles, categories, or other agents that conceptually match its training.
For example, you could create a Mortgage agent and train it with text that is similar to the type of results you expect the agent to return. You can train the agent with text that you type yourself, or with documents. After you train the agent and specify details for it (such as the maximum number of results the agent returns, the minimum conceptual similarity of results and so on), you can run the agent. You can edit or retrain the agent at any time to fine-tune it.
By default agents match against all Content index databases. However, you can restrict the matching to one or more databases.