Amazon Quick Suite for documents in S3: what it indexes, what it costs, and when you don't need it

What Amazon Quick Suite is

Amazon Quick Suite, which AWS also calls Amazon Quick, bundles several things that used to be separate products: a chat assistant, knowledge bases over connected data sources, Quick Sight dashboards, Quick Flows for small automations, Quick Research for longer reports, and Quick Automate for multi-step workflows on the top tier. Amazon Q Business, the previous enterprise assistant, is no longer open to new customers; its documentation now carries a notice recommending Quick for similar capabilities.

For a team whose files are in S3, the relevant part is knowledge bases. A knowledge base is an index built by crawling a data source, and S3 is one of the built-in integrations alongside Google Drive, OneDrive, Confluence, SharePoint and a web crawler. Questions are answered by searching that index, which is what lets Quick answer "which supplier contracts auto-renew?" across thousands of files.

• Chat, knowledge bases, Flows and Research are included on every plan.
• Quick Sight dashboard viewing comes with the Professional and Enterprise plans; creating dashboards is Enterprise only.
• The S3 integration documentation lists an Amazon Quick Enterprise subscription as a prerequisite.

What it costs

The pricing page lists five plans. Free is one user with 1 GB of index. Plus is $20 per user per month billed annually or $25 monthly, up to 300 users, with 10 GB of index per seat. Max is $100 per user per month annually or $125 monthly, with 50 GB per user. Professional is $20 per user per month plus a $250 per account infrastructure fee, with 25 GB of pooled index per user. Enterprise is $40 per user per month plus the same $250 fee, with 50 GB of pooled index per user.

Index storage is measured in gigabytes of raw source file size before Quick processes it, and overage beyond the pooled allocation is $5 per GB per month. Professional includes 8 agent hours a month; Enterprise includes 18 with a $3 per agent hour overage. The pricing page also lists a 30-day Enterprise trial for up to 25 users with the subscription and infrastructure fees waived.

Because the S3 integration requires an Enterprise subscription according to its documentation, the realistic floor for an S3 knowledge base is the Enterprise price. Ten people on Enterprise come to $650 a month before any index overage, and the $250 fee is per account, so a second AWS account means a second fee. Prices change, so check the pricing page before you budget.

How Amazon Bedrock bills document Q&A if you build it yourself

Connecting an S3 bucket

Setup is a console flow under Knowledge. You choose the bucket, either in the same account as your Quick instance or in another AWS account, and supply the bucket path. The bucket must be in the same Region as your Quick Region, and the documentation says cross-account access is also only supported within the same Region. You then choose to add all content or specific S3 prefixes, and the filters are case-sensitive.

Your administrator has to grant Quick access to the bucket first. The troubleshooting page lists the minimum IAM permissions as s3:GetObject, s3:ListBucket, s3:GetBucketLocation, s3:GetObjectVersion and s3:ListBucketVersions. If the bucket is encrypted with a customer managed KMS key, the key policy has to allow both the qbusiness.amazonaws.com and quicksight.amazonaws.com service principals, because Quick uses both during sync.

{
  "Version": "2012-10-17",
  "Statement": [{
    "Sid": "QuickKnowledgeBaseReadOnly",
    "Effect": "Allow",
    "Action": [
      "s3:GetObject",
      "s3:GetObjectVersion",
      "s3:ListBucket",
      "s3:ListBucketVersions",
      "s3:GetBucketLocation"
    ],
    "Resource": [
      "arn:aws:s3:::amzn-s3-demo-bucket",
      "arn:aws:s3:::amzn-s3-demo-bucket/contracts/*"
    ]
  }]
}

S3 bucket policy examples for team file access

The ACL decision you make once

Quick can restrict which users and groups see which documents, but only if you turn on document-level ACLs when you create the knowledge base. The documentation is explicit that the setting cannot be changed afterwards; to switch it on or off later you create a new knowledge base and re-index.

With ACLs enabled, Quick reads either a global acl.json file that maps key prefixes to user and group entries, or per-document metadata files. Each entry names a USER or GROUP and grants ALLOW or DENY. A document with no ACL entry anywhere is not ingested at all, so a bucket where only some prefixes are listed will quietly leave the rest out of the index. Editing the global file means re-indexing the affected prefix, which the best-practices page warns can take hours for very large knowledge bases, and permission changes only take effect at the next sync.

Without ACLs, access is coarse-grained: people either have access to the whole knowledge base or they do not. For a bucket that holds both company-wide policies and board papers, that usually means two knowledge bases rather than one.

• Decide on ACLs before you create the knowledge base; it cannot be toggled later.
• Every document needs an entry, through the global file or a metadata file, or it is skipped.
• Changes to ACLs apply at the next sync, which runs daily by default.

File limits and sync behaviour

Standard text documents such as PDFs, Word files and text files can be up to 500 MB each; tabular files such as CSV and XLS are capped at 50 MB. Whatever the file size, Quick extracts at most 30 MB of text from a single document, and images are limited to 500 per document and 10,000 per GB of index storage. Video and audio are optional and have their own limits.

The knowledge base syncs daily by default and you can change the frequency. Each sync run has a maximum runtime of 14 days, after which it fails with a message asking you to narrow the scope with include and exclude filters or split the content into several knowledge bases. A deletion safeguard lets you set a maximum percentage of documents a sync may remove, so a temporary permissions problem in S3 does not empty the index.

The consequence for a working bucket is that new files are not answerable until the next sync, and the index you pay for is a copy of the extracted text, not the files themselves. The files stay in S3; the knowledge of them lives in Quick.

S3 event notifications: run code when a file is uploaded

When you do not need an index

Most questions people ask about business files are about the one file they have open. What is the notice period in this contract, what does this invoice charge for freight, which version of the policy applies to contractors. For those questions an index is overhead: the file is already known, and what the person wants is an answer that points at the passage it came from.

BucketDesk Document AI takes that route. You open a document in a bucket you have connected, ask in plain language, and a model on Amazon Bedrock answers from that file with a citation that highlights the supporting text. Bedrock runs in your own AWS account through a separate document chat role, so AWS bills the tokens to you at list price and the document is not copied into a third-party index. Documents up to 10 MB are supported, there is nothing to crawl ahead of time, and a file uploaded a minute ago is answerable immediately. Document AI is part of the BucketDesk Ultra plan at $350 a month with 10 operators and unlimited viewers, alongside previews, sharing, uploads and Glacier restore.

The trade is cross-bucket search. BucketDesk will not answer "which of our 4,000 contracts mention a price escalator"; that is what Quick's knowledge bases or Amazon Bedrock Knowledge Bases are for. If that question is a weekly need, budget for an index. If it is occasional, folder search plus per-document questions usually covers it at a fraction of the cost.

Ask questions about S3 documents with Claude on Amazon BedrockAmazon Bedrock Knowledge Bases on S3, or just ask the document?Keep document AI spending under control in your own AWS accountDocument AI on the features page

Side by side

• Where the content lives: Quick keeps an index of extracted text in AWS's Quick service; BucketDesk sends one document per question to Bedrock in your account and keeps no index.
• Freshness: Quick answers from the last sync, daily by default; BucketDesk reads the current object.
• Scope of a question: Quick searches everything in the knowledge base; BucketDesk answers about the open file, with a citation.
• Permissions: Quick needs ACL files or separate knowledge bases; BucketDesk uses the workspace roles and the folder permissions you already set.
• Pricing shape: Quick is per user plus $250 per account plus index overage, with Enterprise required for S3; BucketDesk is a flat plan plus Bedrock tokens on your AWS bill.
• Breadth: Quick adds dashboards, Flows, Research and automation; BucketDesk is a file workspace with previews, sharing and storage operations.