Q: Can AI in space show references for its responses?
A: Yes, as you see in the chat generation, it shows the references from what it’s pulling up. The little footnotes show you exactly where it pulled from the space to power the answer.
Q: Can the chat agent be integrated in Slack?
A: Yes, when setting up a space inside the actions button there is an option to add Slack integration. If your QuickSight admin allows your team and organization to have Slack capability, you can add that as a connector inside any areas of Quick. If added to your space or custom chat, you’ll have access to Slack. Anything in your chat can be sent from Quick to a teammate, yourself, or a channel with any of the answers coming from the chat.
Q: For structured data, if I use integration to bring it in, it will be indexed. I would assume I don’t need to structure data to be indexed. Should I bring structured data into space using actions instead?
A: Depending on where your structured data is coming from, that will determine if you want to use it as a knowledge base or keep it within your QuickSight dashboard and upload it as structured data. If it’s coming from an S3 bucket, you’d put that as a knowledge base into your space. If it’s a small file upload like CSV or Excel format, you could do that from just a file upload. You also have structured data through your dashboard within QuickSight that’s inherited when you uploaded your QuickSight dashboard.
Q: Can you ask a structured question against a CSV file?
A: Yes, the structured data that’s put in the space (like CRM data, customer complaint information, product ratings as CSVs) can be queried. You can ask questions specifically on that structured data and the chat would be able to use the structured data from the space to answer.
Q: Can you generate an agent that explicitly connects to a space?
A: Yes. While the demo didn’t explicitly mention connecting to the space in the prompt, during configuration it automatically added the space as a knowledge base. If it hadn’t done so automatically, you have the option to link a space instead through point and click configuration.
Q: Can emails be sent to non-QuickSight users?
A: Yes. The Outlook integration uploads as your own Outlook account. You just need to have a Quick integration with Outlook to send emails, but the recipients don’t need to have Quick integration. You would just specify the recipient’s email address in the chat.
Q: Have you ever used an agent to help create high charts or optimize existing high chart visuals?
A: This hasn’t been tested yet since Highcharts requires JSON format, but it’s an interesting possibility to explore asking the chat to create Highcharts JSON with specific data.
Q: How can I implement role based security on the data?
A: This will be covered in detail in the admin experience session hosted by Yang on November 14-16. It’s recommended to register for that session for in-depth coverage of admin topics.
Q: What file formats are allowed in spaces import?
A: Spaces can import:
- Unstructured: video, audio, PDF, text, HTML, Word docs, PowerPoint, images
- Structured: CSV, Excel
- Knowledge bases from third party applications
Q: In Q Business, there is a custom connector. In Quick, if I want to use the integration to bring in data from an external data source that is not integrated with Quick, how do I do it?
A: Yes, In Amazon Quick, you can bring in data from external data sources that aren’t natively integrated through the Integrations feature. Unlike Q Business which has custom connectors, Quick uses a different approach with three types of integrations:
- Data Access Integrations
- Establish secure connections to external data sources
- Serve as the foundation for creating knowledge bases
- Cannot be used directly for analysis - you must create a connected knowledge base to make the data accessible
- Action Connectors
- Initiate real-time calls to perform actions in external services
- Used for automating workflows, sending notifications, and triggering processes
- Knowledge Bases
- Connect to data access integrations to make specific content accessible for AI analysis
- Choose which data or knowledge to include from external applications
Q: Is Space capable of eliminating vector embedding and replacing RAG?
A: Amazon Q Spaces are more about organizing and scoping your knowledge for AI interactions rather than fundamentally changing the underlying retrieval technology. They help ensure your AI agents work with the most relevant data for your specific use case, but they still rely on sophisticated retrieval mechanisms (which may include vector embeddings) to find and surface the right information when you ask questions. Think of Spaces as smart containers that help organize your knowledge, while RAG and vector embeddings remain important technologies working behind the scenes to make that knowledge searchable and accessible.
Q: How can I implement role based security on the data?
A: Re. security, all of the Spaces respect the RLS/CLS for your structured data. More reference here: Establishing enterprise governance in Amazon Quick Suite using custom permissions | AWS Business Intelligence Blog
Q: The dashboards my team are using are under wwcsquicksight account, how can I use those dashs in amazonbi account?
A: You would have to recreate the dashboard in the other account.
Q: Are mp4 files enabled in Ireland?
A: Yes, they are believed to be enabled in Ireland.
Q: I see everything when I open the amazonbi Quicksuite but when I switch to the amzlbiaquicksight dashboard I do not see anything and the view is limited, how to resolve this? Am I missing some access requirements?
A: This means the admin of your Quick account needs to enable the AI capabilities.
Q: I have a Dataset that is a query to a table in Redshift. And I have a Dashboard that uses that dataset and has a Table Visual. However the table visual only shows the first 500 rows in the first page, you have to paginate to get the rest of the data. But when I ask my Agent questions about my Space, it only responds for the first 500 rows, it can not access the whole dataset. When I create the Agent or the Space, it only lets me add Dashboards, but I can not add Datasets. Is there a workaround for this? How can I have an Agent that can access the whole dataset?
A: Your admin may have to permission access to unstructured data for you to add other file types but you should be able to add knowledge bases like Redshift without having a cutoff of data.
Q: Can I share the Space with the team and they can also add documents to it to increase the store of knowledge? Does one need to be an owner to be able to upload documents to train the space?
A: Spaces can be shared with teammates as owners to edit the space.
Q: Could you explain the difference between a chat agent and a space?
A: If you want to think of it this way, a space is the contextual knowledge you bring into your agents for more specified context on a specific area of your business. You use a space to power your agents to refine their context in their responses based on your data. Through Quick Automate you can create a multi-agent workflow.
Q: Can Quick Agents integrate (or will integrate in the near future) with the Zero-ETL integrations built in other services?
A: Currently only Quick is where the agents can be used unless with extensions through connectors which will be covered in another session.