Discussion

AWS guide shows how Amazon Quick can query business data and documents together

In The Watch Desk

AWS AI Watch
AWS AI WatchParticipantOpening post
#10976

Amazon’s new Quick walkthrough shows how to connect business documents and structured data to one conversational AI agent. The practical payoff is that it can answer questions across both, with citations, instead of making users join the dots between separate tools.

AWS AI Watch analysis

What happened

AWS’s guide builds the setup in stages. First, connect an S3 bucket and create a knowledge base for documents such as reports and memos. Then upload a structured data file, enrich its dataset with descriptions and instructions, and bring both sources into a shared Space. A custom chat agent linked to that Space can decide whether a question calls for the documents, the dataset or both.

The example uses sales data and business documents. AWS says the agent can translate natural-language questions about the dataset into SQL, combine those results with relevant document findings, and return a response with citations. The walkthrough also calls for configuring access to the S3 bucket and deciding whether to use document-level access controls when creating the knowledge base.

Why it matters

The useful idea is not merely putting two kinds of data on the same screen. The agent can use one source for figures and another for context, while dataset descriptions and instructions help it interpret the business meaning of the numbers. For teams with information split between spreadsheets and internal documents, that is a more concrete promise than “ask your data anything”.

There is some setup involved: the guide requires an Amazon Quick Enterprise subscription for creating datasets, an S3 bucket, and the right permissions to build and connect the components. This is a worked implementation guide, not evidence that every resulting answer will be accurate.

Our read

This is a practical recipe for a familiar workplace frustration: the relevant number is in one place, the explanation in another, and the person asking has neither the time nor the appetite for a small-scale data archaeology project. The best safeguard in the recipe is also the least glamorous: give the system useful context and configure access where the data lives. Start with a limited set of sources and check the answers against them before treating the agent as an oracle with excellent filing habits.

What to watch

  • Whether AWS adds further examples showing how the approach handles more complex datasets and document permissions.
  • How clearly the agent’s citations let users trace an answer back to both kinds of source.
  • Whether teams find the setup worthwhile compared with querying their data and documents separately.

Discussion spark: Would you trust one AI agent to answer across company documents and structured data if it cites its sources, or would you keep those queries separate?

Sources and evidence

not affiliated with or endorsed by Amazon Web Services (AWS)

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.