KIO employs a technique called prompt chaining that enhances the prompt in multiple steps before it presents the final prompt to the agent. In particular, when KIO is asked a question, it will first try to understand the user's intent. Then it checks if there's at least one playbook available that has a description that matches the user's intent, and then injects it into the prompt. This diagram depicts the simplified prompt flow:

Figure 12.15. KIO Prompt Flow