Answers generated from Webex transcript and chat interactions
Q: What’s the accuracy of executive summary?
A: Exec summaries are summarized versions of QS insights. So it uses the QS insights on the visuals and then summarizes that so it keeps accuracy very high but as with any AI generated content, there is a chance of hallucination so should be verified by a human before sharing.
Q: Is there a timeline on when these features will be released to the Ohio region?
A: Hi Joshua, Amazon Bedrock has not yet released support of Ohio so once they do we will be able to plan the work to make it available for genBI as well Supported AWS Regions - Amazon Bedrock
Q/A: I got a good question over DM - sharing answer broadly - Today we released the embedding API/SDK for the genBI Q&A experience.
Embedding the build visual experience is planned for Q3 this year.
We already have edit visual and build calc with natural language features available in embedded authoring. Adding the PM disclaimer that roadmap items are always subject to change but a goal for this year to support build visual in embedding authoring 
Part of why Q only enables a handful of fields to start is to encourage starting with a clear, narrow use case. The more fields enabled, the higher the chance of overlapping language which makes it harder for the model to consistently guess which one the user was thinking of. We do have the “Did you mean” to capture alternate answers to help with valid overlapping cases like “sales in washington” and you have a state field (Washington the state) and city field (washington dc)
So nailing down a narrow case first and slowing adding more fields is a great way to test and build a successful topic.
Q: When was GA and what strategies do you have to deploy this natural language querying outside of quicksight? I have topics used now and it would be nice to embed this functionality somewhere that is in their productivity tools (Slack, Teams) instead of Quicksight?
A: The GA was 4/30/2024!
Today, you can embed this new version of Q. So you can put it in your own website application anything like that but we don’t currently have integration with with Slack or teams.
That is good feedback to know that you are thinking about that.
Q: What is considered a session in the Data Q&A?
A: Taken from the Amazon QuickSight pricing page, a Reader session is defined as:
“a 30-minute interval of time during which a QuickSight Reader can access dashboards and interact with data when subscribed to Capacity pricing. A Reader session starts with user-initiated action (for example, login, dashboard load, page refresh, drill-down, or filtering) and runs for the next 30 minutes. Keeping QuickSight open in a background browser window/tab does not result in active sessions until the Reader initiates action on page.”
Q: Can I move from Author Pro to Author or Reader later ?
A: Yep, exactly. As simple as the admin, an admin or admin Pro can change users to pro or non pro users.
Q: Does Q recognize Click and Clicks as same? or Page and Pages?
A: for plurals, Q should recognize that without having to say both as synonyms.
So say you have “post clicks” as a friendly name, you can put “click” as a synonym and it should also provide answers in Data Q&A if you asked something like “show me clicks for – ad”.
Q: I believe this multi-section /frame answer is new addition, correct? how to turn it off … also if some of these part frames are not relevent to question or not correct, how to deal with it.
A: Yes the multi visual answers are new, you can only turn it “off” by removing q topics but that will also turn off data Q&A in total. If the answers to questions are incorrect, you can report and flag them for review to your admin for review. We always iterate keeping a “human in the loop” as genBI should not always be the only source of truth
Q: If a dataset is filtered by a dataset parameter using direct query , or does the data have to be loaded in “spice” first ?
A: Topics will let you upload any dataset already in your dataset folder in QuickSight, so if that applies to your use case, you are able to use that dataset to create a topic!