Hello! My team is building a chat agent within Quick right now, and has been having much success doing so. Although the visuals provided are insightful, we’d like to look into disabling the visuals that Quick Sight includes, as they are outside of our scope for the time being. We were operating previously with a .csv within a Space, but after some trial and error, we have discovered Datasets and Topics, which has heavily increased the speed of answer for our chat agent. There doesn’t seem to be a very explicit option to remove these visuals from the topic, though. Any help would be appreciated.
Hi @lucasjacob23 and welcome to the Quick Community!
Unfortunately, you are not able disable the visuals that are outputted when you ask a question in Topics. Moreover, even if you technically ask the chat agent the question, if the information is drawing from a topic, the output will have similarities. You can try to adjust the prompt to be more explicit, but I do not that think will work as expected.
Welcome to the Quick Suite community!
Unfortunately, there is no built-in toggle to disable visuals from Amazon Q in Quick Suite Topic responses. Topic-level Custom Instructions are scoped to improving answer accuracy (e.g., business term definitions, field disambiguation) and cannot alter the number or type of visuals returned.
Once an answer is returned, you can only make changes to portions of the interpreted as section to change how the values being used may be aggregated in the visuals.
I’ll mark this as a feature request to promote visibility to the support team.
@Xclipse Thank you for highlighting this. While the visuals are great (and I’m impressed with the speed at which Quick can create random dashboards at will), they’re just not needed by us at this time. Thanks again.
Glad to hear the speed and dashboard capabilities have been impressive so far! Your feedback helps us build a better service. Since this functionality is currently not available, I’ve tagged this as a feature request on your behalf. If you have any follow-up questions, feel free to reach out, otherwise I’ll go ahead and archive/close this case. Thank you!
Hi @lucasjacob23 ,
An update just rolled out and now you’re able to directly connect a dataset to a chat agent, similarly to how you would connect a space or a topic to a chat agent. Ultimately, I think functionality depends on how well your dataset is structured, but I believe this route can be valuable to your use case, with the addition of not having visuals always be included in the output. Try it out and let me know your thoughts!
Hi @lucasjacob23,
Just checking back in since we haven’t heard from you in a bit. I wanted to see if the guidance shared earlier helped resolve your question, or if you found a solution in the meantime.
If you still have any additional questions related to your initial post, feel free to share them. Otherwise, any update you’re able to provide within the next 3 business days would be helpful for the community.
Thank you
Hello @JacobR. Thank you for providing the update. We did see that Quick Suite now supports that functionality, and have been using it ever since. However, my question has now evolved: will Quick’s LLMs ever be able to perform joins without any prep work in the Quick Sight tools? We noticed on testing questions that would normally require a join through two tables sharing a similar primary key that Quick refused to do what we asked since “the query system processes each dataset individually”. This might create problems for our future iterations where we would connect structured tables through Redshift, as opposed to our current approach of uploading tables through .csv files, which are then uploaded as Quick Data Sets. I understand why Quick would work this way, as it’s still obviously dependent on the Topic and Data Set functionality and the data prep tools within each of those platforms. However, is a change to this that would align with our use case on the roadmap for future iterations of the foundation models?
-Lucas
Hi @lucasjacob23 ,
While I cannot definitively speak on any product roadmap or feature rollout, I do want to point out that I know they are trying to revamp the data prep experience in Quick. From what I have seen they are doing small feature rollouts, but I unfortunately do not know much more than that.
Being able to connect/join datasets like that is a cool feature and would be useful. What I can do right now is attach the ‘feature-request’ tag to this post and that can help increase visibility to the Quick team. In the meantime, I would recommend keeping an eye out for any feature announcements on the ‘What’s New’ page (Latest What’s New topics - Amazon Quick Community). If this feature or something similar to this is released, I can try to revisit this topic and let you know.