# Extracting semantic meaning from embedded visual content with Amazon Q Business
When Amazon Q Business processes your input files from a data source, it uses advanced image understanding capabilities to extract semantic information and insights from images and other visuals. For example, it can extract insights from bar charts, technical illustrations, or hierarchical organizational structures.
By extracting semantic meaning from embedded images, Amazon Q Business makes the visual information in your data sources queryable. This makes relevant information easier to find, even if it's conveyed in embedded diagrams or charts. This data provides additional context and nuance to the information in your data sources and builds a more complete knowledge base from your enterprise data.
You can enable content extraction when you add or update a data connector, or when you import a file directly. You can enable it by using the Amazon Q Business console or by using API operations. Processing documents that contain images and visuals takes more time than processing text-only documents.
When you enable content extraction for a connector, we recommend that you use incremental sync mode. If you enable full-sync for your connector, or if you import the same files individually, Amazon Q Business processes the file and corresponding images again. For more information about sync modes in Amazon Q Business, see [Sync mode](connector-concepts.md#connector-sync-mode).
Extracting semantic meaning from embedded images and visuals incurs additional costs. For more information, see [Amazon Q Business pricing.](https://aws.amazon.com/q/business/pricing/)
When Amazon Q Business processes your input files from a data source, it uses advanced image understanding capabilities to extract semantic information and insights from images and other visuals. For example, it can extract insights from bar charts, technical illustrations, or hierarchical organizational structures. This capability now extends to images embedded in web pages, in addition to images in documents like PDFs and presentations.
**Note**
For data sources using web-crawler as the connector, only data sources created after 06/13/2025 will process images both in web-pages and attachments/linked-documents. Data sources created before 06/13/2025, will continue to process images only in attachments/linked-documents (pdf, ppt, word). Please reach out to AWS support for enabling for existing datasources.
**Topics**
+ [End user experience](#semantic-meaning-user-experience)
+ [Guidelines and requirements](semantic-meaning-guidelines-and-requirements.md)
+ [Extracting content from visuals with data connectors](enable-semantic-meanining-data-source.md)
+ [Extracting semantic meaning from audio and video content](Audio-video-extraction.md)
+ [Extracting content from visuals in a file](enable-semantic-meanining-file-upload.md)
+ [Downloading images to add to responses (API operations)](semantic-meaning-adding-img-response.md)
## End user experience
After Amazon Q Business extracts semantic meaning from images and visuals, your end users can ask questions and get answers related to the images. When an end user asks a question, Amazon Q Business retrieves relevant answers from the text and the images. Answers include the images and links for the documents that contain them.
For example, your user might ask, "Can you walk me through the process of enrolling in a health plan on the employee benefits site?" Amazon Q Business would then analyze the company's internal knowledge base, including detailed guides, process documentation, and screen shots showing the step-by-step enrollment process. It would synthesize this information into a clear, easy-to-follow response that outlined each stage of the enrollment journey.
# Guidelines and requirements
The following are guidelines and requirements for extracting content from images:
+ Documents can have up to 3,000 pages.
+ The IAM service role that grants Amazon Q Business permission to access your resources must have `qbusiness:GetMedia` permissions. For a policy example, see [IAM role for an Amazon Q Business web experience using IAM Identity Center](web-experience-iam-role-idc.md)
+ Image extraction is supported for the following image sources: PDF, PPT, Microsoft Word docs, Webpages (for Webcrawler), Google Slides and Google Docs (for Google Drive)
+ You can use the following connectors:
+ [Amazon S3](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/web-experience-iam-role-idc.html)
+ [Google Drive](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/google-connector.html) - Supports embedded images for Google Slides and Google Docs
+ [Confluence (Cloud)](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/confluence-cloud-connector.html)
+ [Confluence (Server/Data Center)](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/confluence-cloud-connector.html)
+ [SharePoint (Online)](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/sharepoint-cloud-connector.html)
+ [SharePoint Server 2019](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/sharepoint-server-2019-connector.html)
+ [SharePoint Server 2016](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/sharepoint-server-2016-connector.html)
+ [Gmail](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/gmail-connector.html)
+ [Microsoft Exchange](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/exchange-connector.html)
+ [OneDrive](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/onedrive-connector.html)
+ [Amazon WorkDocs](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/workdocs-connector.html)
+ [Smartsheet](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/smartsheet-connector.html)
+ [Jira](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/jira-connector.html)
+ [Slack](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/slack-connector.html)
+ [ServiceNow](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/servicenow-connector.html)
+ [Salesforce](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/salesforce-connector.html)
+ [Web Crawler](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/connector-webcrawler.html) - Supports embedded images
+ [Microsoft Teams](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/teams-connector.html)
# Extracting content from visuals with data connectors
You can enable content extraction when you add or update a data connector with the Amazon Q Business console or API operations. Processing images and visuals takes more time than processing text-only for the documents.
## Console
When you add or update a data connector, in the **Multi-media content configuration** section of **Sync scope**, choose **Visual content in documents**. For more information, see the console procedure for your data connector. For example, for a console procedure for Amazon S3, see [Connecting Amazon Q Business to Amazon S3 using the console](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/s3-console.html).
## APIs
To enable content extraction from images when you use the [CreateDataSource](https://docs.aws.amazon.com/amazonq/latest/api-reference/API_CreateDataSource.html) or [UpdateDataSource](https://docs.aws.amazon.com/amazonq/latest/api-reference/API_UpdateDataSource.html) API operations, for `mediaExtractionConfiguration` set `imageExtractionStatus` to ENABLED. The following example shows how to enable content extraction when you create a data source. After you turn on content extraction, you can view media extraction configuration for the data source with the GetDataSource API operation.
```
aws qbusiness create-data-source \
--application-id app-12345abcde \
--index-id index-67890fghij\
--display-name My New S3 Source \
--configuration '{
"S3Configuration": {
"BucketName": "my-s3-bucket",
"DocumentsMetadataConfiguration": {
"S3Prefix": "documents/"
}
}
}' \
--media-extraction-configuration '{
"imageExtractionConfiguration": {
"imageExtractionStatus": "ENABLED"
}
}' \
--description Description of data source connector \
--role-arn arn:aws:iam::123456789012:role/AmazonQServiceRole
```
# Extracting semantic meaning from audio and video content
Amazon Q Business extracts semantic information from audio and video files, making data sources queryable and enhancing information retrieval.
## Capabilities
+ Processes up to 10 GB of video and 2 GB of audio
+ Builds a comprehensive knowledge base
## Enabling Content Extraction
+ Use the Amazon Q Business console or API
+ Available when adding connectors or importing files
## Audio and video extraction
**The System extracts the following from audio and video files:**
+ Transcribed text from audio voice content
+ Time-stamped indexing of content
+ Summary and theme detection
**Important**
Processing multimedia content takes longer than text-only documents. Incremental sync mode is recommended for multimedia processing.
Query audio and video content using natural language, and explore deeper with follow-up questions.
## Processing Considerations
+ Audio and video files take longer to process than text-only files
+ Amazon Q Business doesn't charge for re-synchronizations (re-syncs) when content remains unchanged.
## Additional Information
+ For sync modes, see [Sync mode](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/connector-concepts.html#connector-sync-mode)
+ For pricing details, see [Amazon Q Business pricing](https://aws.amazon.com/q/business/pricing/)
## Guidelines and requirements
**The following guidelines and best practices apply to content extraction from audio and video. Supported formats:**
+ Audio files - MP3, WAV, M4A, FLAC, OGG
+ Video files - MP4, MOV, M4V
**File limitations:**
+ Audio and video maximum duration: 4 hours
+ English-language content only
**The following connectors use incremental sync mode for connectors processing multimedia content, and can consider additional costs for multimedia processing:**
+ [ Amazon S3](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/s3-connector.html)
+ [Google Drive](https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/google-connector.html)
## Ingest Audio/Video documents using the console
1. Login to your personal AWS account
1. Create a new application or use an existing application
### Ingest Audio/Video files using direct upload
Open the application, whether it is a newly created application or an application you have used previously:
1. Click **Add data source**
1. Select the **Upload Files** option
1. Upload a supported Audio/Video file
1. In the **File Indexing Configuration**, select the audio/video files option based on the files you uploaded
1. Click the upload button on the bottom right corner
1. Monitor the uploaded file source under the data source tab. The status will update in sequence: "Received" => "Processing" => "Indexed/Updated"
1. Once the status is Indexed/Updated, it means your file is indexed successfully
**Note**
For file uploads, the Audio/Video size limit is consistent with the general file size limit of 50 MB.
### Ingest Audio/Video documents using data source
Open the application, whether it is a newly created application or an application you have used previously:
1. Click **Add data source**
1. Select any supported data source
1. Fill in the following information:
1. Enter the data source name
1. For the IAM role, select from the dropdown menu **Create new service role **
1. Enter the S3 bucket containing the Audio/Video files in the **Data Source Location** field under **Sync scope**
1. Set **Sync Mode ** and **Sync Run **as per your requirement
1. Click on **Add data source** on the bottom right corner
1. Navigate to the **Data sources** pag and select the data source created and click **Sync now**
1. Click on the data source created, wait for the sync to complete and verify no documents from the Sync history tab **Failed to be indexed**
# Extracting content from visuals in a file
You can enable content extraction from a file with the Amazon Q Business console or API operations. Processing images and visuals takes more time than processing text-only for the documents.
## Console
When you upload documents directly to an Amazon Q Business application environment, in the **Multi-media content configuration** section of **Select files**, choose the **Visual content in documents** option. For step by step instructions, see [Uploading files](upload-docs.md).
## APIs
To enable content extraction from a file when you use the [BatchPutDocument](https://docs.aws.amazon.com/amazonq/latest/api-reference/API_BatchPutDocument.html) API operation, in the `ImageExtractionConfiguration` you set the `imageExtractionStatus` to `ENABLED`.
```
aws qbusiness batch-put-document \
--application-id app-12345abcde \
--index-id index-67890fghij \
--role-arn arn:aws:iam::123456789012:role/ServiceRoleName \
--documents '[{
"Id": "doc1",
"MediaExtractionConfiguration": {
"ImageExtractionConfiguration": {
"ImageExtractionStatus": "ENABLED"
}
}
}]'
--data-source-sync-id sync-12345
```
# Downloading images to add to responses (API operations)
If you have implemented your own application with the Amazon Q Business Chat and ChatSync APIs, you can use the [GetMedia](https://docs.aws.amazon.com/amazonq/latest/api-reference/API_GetMedia.html) API operation to download the images to add to chat responses. You can find the `mediaId` using the Chat, ChatSync and ListMessages API operations. The `mediaId` is listed in the `textMessageSegments` as part of the source attribution.
**Note**
The `mediaBytes` field in the GetMedia API response contains binary image data that may not require base64 decoding, depending on your implementation.
```
aws qbusiness get-media \
--application-id app-12345abcde \
--conversation-id conv-67890fghij \
--media-id media-12345abcde \
--message-id msg-67890fghij
```