How Agents Work

In the simplest format, an agent automatically performs a specified catalog analysis based on a defined schedule, and examines the results for a specific problem or opportunity. If the specific problem or opportunity is detected in the results, then an alert is generated and delivered to specified recipients and to subscribers to the agent, using the delivery options that are specified for each person.

For more information on alerts, see "What Are Alerts?" For more information on delivery options, see "What Are Devices and Delivery Profiles?"

To handle more complex requirements, agents can invoke actions that trigger other agents, scripts, Java programs, or applications. Results can be passed between agents, and to other applications or services through XML, HTML, or plain text. For example, an agent might run an analysis to identify all current product orders over a specified dollar amount that cannot be filled from a regional warehouse. The results can be passed to another agent that runs an analysis to locate alternative sources for these products. A final agent might be triggered to feed information into a corporate CRM system and to notify the appropriate account representatives of the alternative sourcing.

You might automatically be a recipient of alerts generated by some agents, and agents created by others might be available for you to subscribe to. You can also create your own agents if you have the appropriate permissions and responsibilities. Depending on the level of authority that you have, you can selectively share agents with others or make agents available for all users.