Asset Enrichment: automatic metadata enrichment with AI
With the Asset Enrichment automation, every image uploaded to THRON automatically receives a description, alt text, and semantic attributes — without any manual intervention. The result: the catalog is searchable and accessible right from the moment of upload, and your team stops wasting time on manual tagging.
Prerequisites
Before configuring the automation, verify that the fields you intend to populate already exist in the attributes of your digital assets. The automation does not create new attributes: it only populates existing ones.
The correct attribute type is required for each content you want to generate:
Content to generate |
Required attribute type |
|---|---|
Description |
Long Text |
Alt text |
Short Text |
Dominant color |
Simple Select |
Orientation |
Simple Select |
Content type |
Simple Select |
Keywords |
Long Text |
Configuration Guide
Step 1 — Access and create the automation
Go to Platform Settings → Flows & Automations → Automations
Click on New and select Asset Enrichment
Step 2 — Configure description and alt text
Description
The description is generated and updated automatically every time the master image is uploaded or modified, and written in the Long Text attribute that you select.
Configure the following parameters:
Parameter |
Description |
|---|---|
Length |
Defines the length of the generated description |
Tone of voice |
Text register: neutral, marketing, technical, institutional |
Audience (optional) |
Recipient of the description (e.g., B2B buyer, end consumer) |
Additional information (optional) |
Additional instructions to guide generation (e.g., usage context, specific focus) |
Note: The parameters apply to all subsequent executions. Changes to the configuration do not have a retroactive effect on already processed assets.
Alt text
Alt text is always active and requires no prompt configuration. Simply select the Short Text attribute where it should be written.
Maximum length: 125 characters
It is not affected by tone of voice or audience parameters
It is visible in the player as accessibility information
Step 3 — Map the attributes to be enriched
For each attribute, map the automation label to the corresponding attribute in your assets. Only activate the attributes you actually want to populate.
Attribute |
Required type |
What it populates |
|---|---|---|
Dominant color |
Simple Select |
The visually prevalent color in the image |
Orientation |
Simple Select |
Aspect ratio: vertical, horizontal, square |
Content type |
Simple Select |
Material type: product image, lifestyle, packaging… |
Keywords |
Long Text |
Comma-separated semantic keywords |
Two important things:
For Simple Select attributes, the automation chooses exclusively from the options already present in your attribute schema. Use the Info button to view them directly in the interface.
For Keywords, also specify the purpose — the goal for which they are generated (e.g., SEO, internal search, classification) — to guide the selection of the most relevant terms.
Warning: If you upload assets before completing the mapping, the automation flags the configuration as incomplete. Assets uploaded in this state are not retroactively enriched.
Step 4 — Set the languages
Select the primary language in which the description must be generated
Add any additional translation languages
The automation performs the operations in this order:
Generation of the description in the primary language
Translation into the selected languages
Generation of consistent alt text for each language
Classification of semantic attribute options
Classification of keywords in the various languages
Step 5 — Choose the overwrite behavior
Define what the automation should do when it finds an asset with already populated attributes:
Option |
Behavior |
|---|---|
Overwrite attributes if populated: ON |
The automation also populates already filled attributes |
Overwrite attributes if populated: OFF |
The automation does not intervene on attributes already present |
Recommended Configuration
If this is the first time you are configuring the automation, start with this preset:
Parameter |
Recommended value |
Why |
|---|---|---|
Tone of voice |
Neutral |
Works for all contexts, changeable later |
Description attribute |
Description (Long Text) |
Field already present by default |
Alt text attribute |
Short Text field dedicated to accessibility |
— |
Active semantic attributes |
Orientation + Content Type |
Useful right away for filtering the catalog |
Overwrite |
OFF |
Protects manually entered metadata |
Execution report
At the end of each batch, THRON automatically generates a report with the processing outcome for each processed asset.
Field |
Description |
|---|---|
filename |
Name of the processed file |
content_id |
Unique identifier of the asset |
description |
Description generated by AI |
alt_text |
Alt text generated by AI |
keywords |
Assigned keywords |
dominant_color |
Assigned value |
orientation |
Assigned value |
content_type |
Assigned value |
error |
Any error encountered |