You can now choose how deeply your Knowledge Base searches for relevant information with two new search modes:
Fast selects what it thinks are the most relevant Knowledge Base chunks, and then searches this subset of chunks to pull information to respond to users. This prioritizes speed and low search overhead.
Thorough searches every chunk in your Knowledge Base before selecting the most relevant information, helping uncover additional useful context.
A couple things to keep in mind:
• Thorough mode may take longer because it searches across your entire Knowledge Base.
• The maximum number of chunks ultimately pulled into a response is still determined by your token limits, which can be found in the Configure tab of your agent builder.
Whether you want a quick answer or the deepest possible search, you now have more control over how your Knowledge Base works. Give both modes a try and let us know what you think! 
