DocumentationAEMUser Guide

[AEM Assets]{class="badge positive" title="Applies to AEM Assets)."}

Smart Tags for AEM Assets using-smart-tags

Last update: September 25, 2026
  • Applies to:
  • Experience Manager as a Cloud Service

CREATED FOR:

  • Admin
  • User

Smart Tags in Adobe Experience Manager (AEM) Assets are keywords that not only appear in an asset but also best describe it, automatically generated by artificial intelligence to make digital content easier to find. Organizations possess vast and rapidly expanding libraries of digital assets, and this growth makes efficient discovery essential to productivity. Locating a specific asset within such a large collection is one of the most time-consuming obstacles in digital asset management. To solve this, metadata and tags enhance the searchability of digital assets. Organizations use taxonomy-controlled vocabularies in asset metadata—curated keyword lists that employees, partners, and customers commonly use to refer to and locate digital assets.

Why Tagging Matters

Tagging assets with a taxonomy-controlled vocabulary ensures they can be quickly identified and retrieved through search. Consider how words arranged alphabetically in a dictionary are far easier to find than words scattered at random. Tagging serves the same purpose: it organizes assets according to business taxonomy because it aligns each asset with the controlled vocabulary your organization already uses, ensuring the most relevant results surface first. As a result, teams spend less time hunting for files and more time using them.

Smart Tags improve relevance and reduce friction in everyday search:

  • Taxonomy alignment: Assets are labeled with the model names, product terms, and categories your business relies on, so only relevant results appear.
  • Variation tolerance: Whether a user searches for “runners” or “running shoes,” they do not need to worry about typos, spelling variations, or alternate search terms—Smart Tags recognize them all.
  • Faster campaign workflows: For example, a car manufacturer can tag car images with model names, so that when designing a promotional campaign, only the relevant images are displayed—eliminating manual filtering and speeding up creative production.

This automated, consistent labeling helps large teams maintain a shared, searchable vocabulary across thousands of assets without relying on manual, ad hoc keywording.

How Smart Tags Work

In the background, this functionality uses the artificial intelligence framework of Adobe AI. By default, Adobe AI automatically applies Smart Tags to uploaded assets, generating text aligned to the business taxonomy so that newly ingested content becomes discoverable immediately, without additional manual effort.

Prerequisites and configuration smart-tags-prereqs-config

Smart Tags is automatically provisioned for Adobe Experience Manager (AEM) as a Cloud Service. As a result, no manual setup or configuration is required to begin using the feature.

Because provisioning is handled automatically as part of the Cloud Service environment, administrators do not need to install additional components, enable services individually, or complete any preliminary configuration steps before Smart Tags becomes available. This automatic provisioning is a defining characteristic of the Cloud Service delivery model, in which capabilities are delivered ready to use rather than requiring on-premise style installation and setup.

Key points:

  • No configuration required: Smart Tags is provisioned out of the box, so it is ready for use immediately within AEM as a Cloud Service.
  • Automatic provisioning: The feature is enabled by default as part of the Cloud Service environment, removing setup overhead for administrators.
  • No prerequisites to satisfy: Because the capability is delivered automatically, there are no additional prerequisites that must be met before you can begin working with Smart Tags.

This means users can proceed directly to applying and working with Smart Tags without any preparatory administrative tasks, streamlining the path to leveraging the feature within AEM as a Cloud Service.

Smart Tags workflow smart-tags-workflow

Adobe AI powered smart tagging uses artificial intelligence (AI) models to analyze content and automatically add tags to assets. This reduces the time required for Digital Asset Management (DAM) users to deliver rich experiences to their customers, because relevant tags are generated automatically rather than added manually. The Smart Tags are displayed in descending order of their confidence score in asset properties, so the most relevant tags appear first.

  • Image-based assets
    For images, the Smart Tags are based on visual aspects of the content. Images in many formats are tagged using smart content services. Smart Tags are applied to the supported file types that generate renditions in JPG and PNG format.

  • Video-based assets
    For video-based assets, Adobe Experience Manager as a Cloud Service enables video tagging by default. As is the case with image and text based tags, videos are also auto-tagged when you upload new videos or reprocess existing ones. Adobe AI generates two distinct sets of tags for each video: one set corresponds to objects, scenes, and attributes in that video, and the other set relates to actions such as drinking, running, and jogging. Also check opt out video smart tagging.

  • Text-based assets
    For supported assets, Experience Manager already extracts the text, which is then indexed and is used to search for the assets. However, Smart Tags based on keywords in the text provide a dedicated, structured, and higher priority search facet. Keyword-based Smart Tags improve asset discovery compared to a standard search index, because they surface assets as a dedicated, structured, higher-priority facet rather than relying on plain indexed text.

    Smart-tag-types

    For text-based assets, the efficacy of Smart Tags does not depend on the amount of text in the asset but on the relevant keywords or entities present in the text of the asset.

Smart Tags are implemented in AEM Assets using the following workflow:

  1. Create or upload an asset in AEM. Out of the box tags are generated for image, video, and text based assets.

  2. If you find that specific tags are not generated, then you can train your image-type tags accordingly. Refer to Smart Tags training.

Supported file formats for Smart Tags supported-file-formats

Smart Tags supports three categories of assets for automatic tagging: images, text-based assets, and video assets. Each supported format below can be ingested and analyzed by Smart Tags to generate metadata automatically, which is why using a format from the supported list is essential for successful tagging. Image files are identified by their Multipurpose Internet Mail Extensions (MIME) types, text-based assets are identified by file format, and video assets are supported based on both file format and the underlying codec (such as H264/AVC).

Images (MIME types)
Text-based assets (file formats)
Video assets (file formats and codecs)
image/jpeg
CSV
MP4 (H264/AVC)
image/tiff
DOC
MKV (H264/AVC)
image/png
DOCX
MOV (H264/AVC, Motion JPEG)
image/bmp
HTML
AVI (indeo4)
image/gif
PDF
FLV (H264/AVC, vp6f)
image/pjpeg
PPT
WMV (WMV2)
image/x-portable-anymap
PPTX
image/x-portable-bitmap
RTF
image/x-portable-graymap
SRT
image/x-portable-pixmap
TXT
image/x-rgb
VTT
image/x-xbitmap
image/x-xpixmap
image/x-icon
image/photoshop
image/x-photoshop
image/psd
image/vnd.adobe.photoshop

The image category covers standard web and print formats — including JPEG, PNG, GIF, TIFF, and BMP — as well as portable pixmap variants and Adobe Photoshop (PSD) formats. Text-based assets span common document types such as PDF, DOCX, and PPTX, along with subtitle and caption files (SRT and VTT). Video assets include widely used containers such as MP4, MKV, MOV, AVI, FLV, and WMV, each supported through the specific codecs listed alongside them.

Preparing an asset for out of the box smart tagging

Smart Tags are automatically generated by Adobe AI when you upload assets to Adobe Experience Manager as a Cloud Service. Smart Tags are descriptive metadata keywords—applied without manual effort—that identify the visual content, subjects, and concepts within an asset, making assets easier to search, filter, and organize across your digital asset management library.

How out-of-the-box smart tagging works

Smart Tags are added to videos and seen in Basic tab of asset Properties

The tagging workflow runs automatically as part of asset processing:

  1. Upload the asset to Adobe Experience Manager as a Cloud Service.
  2. Processing runs automatically through Asset microservices, which analyze the uploaded content.
  3. Adobe AI generates Smart Tags and applies them to the asset. Asset microservices uses Adobe AI to interpret the asset and produce these Smart Tags, so relevant keywords are added without human intervention.
  4. View the results on the Basic tab of the asset Properties page, where the generated tags appear under Smart Tags.

Because these tags are derived directly from the content of each asset, they streamline downstream tasks such as search relevance, content discovery, and metadata consistency across large asset volumes.

IMPORTANT
Review these automatically generated tags to ensure that they conform to your brand and its values. AI-generated tagging accelerates metadata creation, but a review step confirms that the applied keywords accurately reflect your intended messaging and remain aligned with your brand standards before the assets are used or distributed.

Untagged Assets in DAM smart-tag-existing-assets

Existing or older assets in Digital Asset Management (DAM) are not smart tagged automatically. Smart tagging applies during asset processing, so assets that were ingested before smart tagging was enabled will not carry Smart Tags until they are processed again. Administrators must manually Reprocess these assets to generate Smart Tags for them, because the tagging workflow runs as part of the reprocessing job rather than retroactively.

To generate and view Smart Tags for existing assets, follow these steps:

  1. Reprocess the assets manually. Trigger Reprocess on the existing assets so the smart tagging workflow analyzes them and generates tags.
  2. Wait for the process to complete. Once reprocessing finishes, the generated Smart Tags are written to each asset’s metadata.
  3. Open the asset properties. Navigate to the Properties page of any asset within the folder.
  4. Locate the Smart Tags. The automatically added tags appear in the Smart Tags section of the Basic tab.

Because reprocessing regenerates metadata, the applied Smart Tags are sorted in descending order of confidence score. As a result, the tags the system is most confident about appear first, making it easier to evaluate the most relevant classifications at a glance.

Confidence Score confidence-score

Asset search results are ranked by confidence score, a numerical measure of how reliably a tag applies to an asset. The confidence score improves search relevance beyond what an inspection of the assigned tags of any asset suggests, because it prioritizes the tags most likely to be correct rather than treating every tag equally.

How Confidence Scores Rank Results

Each Smart Tag carries a confidence value that reflects the system’s certainty that the tag accurately describes the asset. Higher-confidence tags surface first, ensuring that the most relevant and reliable matches appear at the top of the results.

Because inaccurate tags typically receive low confidence scores, they rarely appear at the top of the Smart Tags list for assets. As a result, low-confidence, potentially mismatched tags are pushed down the ranking, which reduces noise and helps users locate the correct assets more quickly.

NOTE
Manual tags are assigned a confidence of 100% (maximum confidence). Therefore, when assets carry manual tags that match the search query, those assets are displayed before Smart Tags matching the same query. This ordering reflects the higher trust placed in human-assigned tags: because a person deliberately applied the tag, the system treats the match as fully reliable and prioritizes it over automatically generated Smart Tags.

Moderate Smart tags moderate-smart-tags

Adobe Experience Manager as a Cloud Service allows you to moderate Smart Tags to improve asset discoverability. Curating Smart Tags lets you:

  • Remove inaccurate tags assigned to your brand assets, keeping tag metadata clean and trustworthy.

  • Refine tag-based searches for assets by ensuring that your asset appears in search results for the most relevant tags. Refining tags therefore eliminates the chance of unrelated assets appearing in search results, so users retrieve only the assets that match their intent.

  • Assign a higher rank to a tag to increase its relevance to an asset. Promoting a tag increases the chances of that asset appearing in search results when users search on that tag, because higher-ranked tags carry greater weight in tag-based retrieval.

Moderate video Smart Tags

To know more about how to moderate the Smart Tags for assets, see Manage Smart Tags.

NOTE
Any tags moderated using the steps in Manage Smart Tags are not retained when the asset is reprocessed. On reprocessing, the original sets of tags are displayed again.

Manage Smart Tags and asset searches manage-smart-tags-and-searches

Smart Tags are the metadata labels automatically applied to your digital assets to describe their content and make them discoverable in searches. Curating these Smart Tags lets you remove any inaccurate tags automatically assigned to your brand assets, so that only the most relevant tags remain visible. Effective moderation of Smart Tags serves two core goals: keeping the tags attached to each asset accurate, and ensuring assets surface for the searches most likely to match them.

Moderating Smart Tags refines tag-based search results by ensuring that your assets appear for the most relevant tags. This works because search systems match a query against the tags assigned to each asset — when the tag set is clean and accurate, only the genuinely relevant assets are returned. As a result, moderation eliminates the chances of unrelated assets showing up in search results, which reduces search clutter and helps users locate the correct asset faster.

You can also assign a higher rank to a tag to increase that tag’s relevance for the asset. Promoting a tag is useful when a particular tag best represents how users search for an asset, because ranking pushes the most important descriptors to the top. As a result, promoting a tag for an asset directly increases the likelihood of the asset appearing in search results when a search is performed on that particular tag.

To moderate the Smart Tags of your digital assets:

  1. In the search field, search for digital assets based on a tag.

  2. Inspect the search results to identify the digital assets that are not relevant to your search.

  3. Select an asset, and then select Manage tags icon from the toolbar.

  4. From the Manage Tags page, inspect the tags. If you do not want the asset to be searched based on a specific tag, then select the tag and select Delete icon from the toolbar. Alternatively, select close icon next to the label. Removing an irrelevant tag prevents the asset from being returned for that tag in future searches.

  5. To assign a higher rank to a tag, select the tag and select Promote icon from the toolbar. The tag you promote is moved to the Tags section.

  6. Select Save and then select OK to close the Success dialog.

  7. Navigate to the Properties page for the asset. Observe that the tag you promoted is assigned a high relevance and, therefore, appears higher in the search results.

By default, Experience Manager combines search terms with an AND or OR clause to find any of the search terms within the applied Smart Tags. Applying Smart Tags does not change this default matching behavior. For example, when searching for woman running, assets that contain only the woman keyword or only the running keyword in their metadata do not appear in the results by default. However, an asset that has been smart tagged with either woman or running does appear for the same query, because Smart Tags extend the reach of the search beyond literal metadata.

As a result, Experience Manager returns a combination of two distinct match types:

  • Assets that contain both the woman and running keywords in the metadata.

  • Assets that have been smart tagged with either of the keywords.

Experience Manager displays search results that match all search terms in the metadata fields first, because exact metadata matches indicate the strongest relevance, followed by results that match any of the search terms in the Smart Tags. In the example above, Experience Manager orders the search results approximately as follows:

  1. Matches of woman running across the various metadata fields.
  2. Matches of woman running in Smart Tags.
  3. Matches of woman or of running in Smart Tags.

Opt out of smart tagging opt-out-smart-tagging

You can opt out of smart tagging on upload at the folder level to expedite asset processing. Because automated tagging of assets runs in parallel with other asset processing tasks such as thumbnail creation and metadata extraction, it can be time consuming, and disabling it removes that additional processing overhead. Opting out at the folder level ensures that assets uploaded to that specific folder skip automated Smart Tags generation, allowing the remaining processing tasks to complete faster.

To opt out of automated Smart Tags generation for assets uploaded to a specific folder:

  1. Open the Asset Processing tab in the folder Properties.

  2. In the Smart Tags for Videos menu, for example, the Inherited option is selected by default and video smart tag is enabled.

    Disable smart tagging

    When the Inherited option is selected, the inherited folder path is also visible, along with the information indicating whether the path is set to Enable or Disable. This lets you confirm exactly which parent-folder setting the current folder is inheriting before you override it.

  3. Select Disable to opt out of smart tagging for assets uploaded to the folder.

  4. You can similarly opt out of smart tagging for the other asset types: Smart Tags for Text, Smart Tags for Image, and Color Tags for Images. Each type is controlled independently, so you can disable tagging for some asset types while leaving it enabled for others.

IMPORTANT
If you have opted out of tagging on a folder at the time of upload and want to smart tag the assets after upload, then Enable Smart Tags from the Asset Processing tab of the folder Properties and use the Reprocess Asset option to add Smart Tags to the assets. This ensures previously skipped assets receive Smart Tags without requiring a re-upload.

The Smart Tags models automatically identify and apply tags to assets, but the current version operates within defined technical boundaries. Understanding these constraints, and applying the recommended search practices, ensures reliable tagging and retrieval.

Known Limitations

The Smart Tags models rely on visual recognition, and like most automated image-analysis systems, they perform best on clearly distinguishable visual features. The current version has the following limitations:

  • Cannot recognize subtle differences in images. For example, the models do not reliably distinguish slim-fit versus regular-fit shirts.
  • Cannot identify tags based on tiny patterns or small portions of an image. For example, small logos printed on shirts fall below the threshold of reliable detection.
  • Does not handle non-visual, abstract, or subjective aspects. This includes the year or season of release of a product, the mood or emotion evoked by an image, and the subjective connotation of a video — because these attributes are not directly observable in the visual content itself.
  • Does not resolve fine visual differences in products. For example, the models do not consistently separate shirts with collars from shirts without collars, and they do not detect small product logos embedded on products. Both cases involve minute visual distinctions that automated recognition finds difficult to isolate.
  • Auto-tags only videos smaller than 300 MB in file size. The Adobe artificial intelligence (AI) service skips video files larger than 300 MB to keep processing efficient and within service limits, so larger videos are not auto-tagged.

Best Practices for Searching Smart Tags

  • To search for files with Smart Tags (regular or enhanced), use the Assets search (full-text search). There is no separate search predicate dedicated to Smart Tags, so full-text search is the supported method for locating tagged assets.
  • Compared with general tags, assets tagged using business taxonomy are easier to identify and retrieve through tag-based searches. This is because business taxonomy applies structured, domain-specific terms that align more precisely with how users search, resulting in more accurate and consistent retrieval.

Frequently asked questions faq-smart-tags

How do Smart Tags improve search experience of an asset?
Adobe AI tags the assets automatically once you upload them. The automated process runs so fast at the backend that you will see tags added to your assets within a few seconds after the upload is completed. This means assets become searchable by their auto-generated tags almost immediately, without any manual tagging effort.
What happens if the Smart Tags list is inaccurate or showing unwanted tag?
You can remove any inaccurate or unwanted tag from the list. For example, an automobile dealer removes the “damaged” tag from the list to keep vehicle assets from surfacing under that term in search results.
How can you prioritize assets containing same tags?
Yes, you can prioritize assets containing the same tags. Promote a tag into the Smart Tags list of an asset to perform prioritization. Promoting a tag prioritizes the images that appear in the search results for that particular tag, so the most relevant assets rank higher for that keyword.
Is the application of Smart Tags limited to a particular folder?
No. Smart Tags are configurable and can be applied to any folder inside the Digital Asset Management (DAM) repository.
How may I know that tagging needs training?
Refer to Determining the requirement of Smart Tags training.
What are the supported file formats for tagging an asset?
Refer to Supported file formats.
In which language smart tags are generated?
Smart tags are generated in English only. To make tags available in other languages, translate the entire asset, including its metadata, because the tags are stored as part of the asset’s metadata.
I do not want to use Smart Tagging anymore.
You can opt out of Smart Tagging at any time to discontinue automatic tagging for your assets.

Troubleshooting Scenarios troubleshooting-scenarios

Feature naming and navigation feature-naming-and-navigation

AI-Generated metadata, also called Smart Tags, automatically tags assets using artificial intelligence. This capability is exposed through two related but distinct surfaces: a platform-level configuration toggle that turns the enhancement on or off, and a separate training workspace used to build a custom Smart Tag model. Because these two surfaces live in different parts of the interface, they are reached through different navigation paths, summarized below.

Task
Correct location
Frequently Encountered Issues
Enable or disable the AI-Generated metadata (Smart Tag enhancements) at the platform level
Tools > Assets > Assets Configurations > Smart Tag Enhancement Configuration
This is a configuration toggle, self-service.
Train a custom Smart Tag model
Assets > Smart Tag Training, reached from /aem/start.html
This feature is not available under Tools or under Tools > Assets > Smart Tag Training. Custom model training is administered from the start console rather than the Tools configuration path, so looking for it under Tools leads to a dead end.

Troubleshooting: Smart Tag Training not visible

If Smart Tag Training is not visible under Assets in /aem/start.html, the cause is typically a permissions, feature-availability, or asset-support gap. Check the following, in this order, because each step must be satisfied before the option appears:

  1. User group or role assignment — confirm the signed-in user belongs to a group or role that grants access to Smart Tag Training.
  2. Auto-tagging enablement — verify that Smart Tags auto-tagging is enabled for the instance.
  3. AEM as a Cloud Service availability — confirm the required Adobe Experience Manager (AEM) as a Cloud Service version or feature is available for the environment.
  4. Supported asset types — verify that the asset types in question are supported for custom model training.

For more information on AI-Generated metadata or Smart Tags, see Enhancing content discovery with AI-Generated metadata.

Enablement history and self-service status enablement-history-and-self-service-status

AI-generated metadata enablement and disablement is now a fully self-service capability. Previously, enabling or disabling this feature required Adobe Engineering intervention in some cases; that dependency has been eliminated. As a result, administrators gain direct, on-demand control over the feature without relying on the Adobe Engineering team to process the change.

Administrators can enable or disable AI-generated metadata at any time directly through the Smart Tag Enhancement Configuration screen. Because the change is handled entirely within the configuration screen, there is no need to open a support ticket for enablement, which removes the turnaround delay previously associated with engineering-assisted requests.

What the self-service model means in practice:

  • Immediate control: Enablement and disablement take effect from the Smart Tag Enhancement Configuration screen without waiting on a support or engineering request.
  • No support ticket required: The previous requirement to raise a support ticket for enablement has been removed, streamlining the workflow.
  • Anytime toggling: Administrators can turn the AI-generated metadata feature on or off whenever needed, providing flexibility to adjust the configuration as requirements change.

This shift to a fully self-service model reduces reliance on Adobe Engineering, shortens the time required to activate or deactivate AI-generated metadata, and gives administrators direct ownership of the feature’s status.

Requirements and Limitations requirements-and-limitations

  • Add-on requirement: Smart Tags requires the Smart Content Services add-on or entitlement. Without this add-on or entitlement, Smart Tags functionality is unavailable.
  • Data residency: Smart Tags and Dynamic Media both process and deliver content through Adobe’s global cloud infrastructure. For customers with data residency requirements (for example, data must stay within a specific country), be explicit that using these features may involve processing or transferring asset data outside that region. This is a characteristic of the global infrastructure, not a configurable regional restriction. As a result, organizations in regulated industries or jurisdictions with strict data-localization rules should account for this cross-region processing during their compliance assessments, because the behavior cannot be constrained to a specific geography through AEM settings.
  • Model selection and provenance: Image generation through Smart Tags or AI features in Adobe Experience Manager (AEM) Assets uses Adobe-owned (first-party) Adobe Firefly models such as Image Model 4 only by default. There is currently no AEM-side configuration or model picker to select or restrict which model is used, and model attribution is not exposed in asset metadata. As a result, compliance reporting must be planned externally when model provenance needs to be tracked, because AEM does not surface this attribution automatically. Teams that require a documented audit trail of which model produced a given asset should therefore maintain that record outside AEM.
  • Generate Variations licensing: The separate Generate Variations generative feature is licensed specifically under AEM Sites with Edge Delivery Services (EDS) and is not available to Assets-only environments regardless of Smart Tags or AI-Generated metadata status. Do not conflate the two generative AI capabilities, because they are licensed, provisioned, and scoped independently — enabling Smart Tags in an Assets environment does not grant access to Generate Variations.

Resolution Workflow resolution-workflow

  1. Smart Tag Training menu option is not visible: Confirm you are looking under Assets > Smart Tag Training in /aem/start.html, and not under Tools. The menu path determines visibility, so navigating to the incorrect parent menu is the most common reason the option cannot be found.
  2. Feature appears disabled and the toggle cannot be located: Check Tools > Assets > Assets Configurations > Smart Tag Enhancement Configuration. This configuration is self-service and does not require an Adobe support ticket, because the toggle is administrator-controlled within the Assets configuration area.
  3. Data residency or compliance question: Clarify that Smart Tags and Dynamic Media use Adobe’s global infrastructure. As a result, asset processing can occur outside your home region, which is an important consideration for data residency and compliance reviews.
  4. Compliance question about which AI model was used: Clarify that only Adobe-owned Firefly models — Adobe’s proprietary generative AI models — are invoked. There is no selectable alternative model, and model attribution is not currently surfaced in metadata.
  5. Generate Variations are not available: Confirm the organisation has the AEM Sites and Edge Delivery Services (EDS) stock-keeping unit (SKU). This is a separate entitlement from Smart Tags or AI-Generated metadata, so provisioning one does not automatically grant access to the other.

See also

recommendation-more-help
experience-manager-cloud-service-help-main-toc