Answers generated from Webex transcript and chat interactions
Q: Can you show how you got to the story section again?
A: When you are on the landing page of your QuickSight dashboard, there is a dropdown tab on the top right of your dashboard’s screen named “build” with a small Amazon Q logo. If you click on that, there should be two dropdown option for an executive summary and one for data story. Click the data story option to enter the data story build screen.
Q pt.1: Is the feedback also trained with Bedrock? ie. if I report a wrong answer explaining, will Q learn how to answer correctly?
A pt.1: Yes, Amazon Q is powered by Bedrock and our in-house models. If an answer is incorrect, it will be sent to the author who built the dashboard and the Q topic, which refines the answers being outputted/ can be edited to be more accurate. Think of it like a feedback loop to your dashboard builder to help Q learn to provide better answers.
Q pt.2: Ok so Q will not learn by himself, the author needs to change / refine the model
A pt.2: We keep a huamn-in-the-loop as to prevent hallucinations in understanding your data.
Q: Are we going to have a session on how to better refine the answers? I am struggling with that
A: That’s a great suggestion, we will put that in future interested topics as many customers have also brought that up.