Overview

Natural Language Search (NLS) is an AI-assisted search method. It enables users to enter queries in everyday language rather than specific keyword matches or complex Boolean operators. This method helps the search engine understand the meaning of search queries semantically and uses existing asset information. This process produces better, broader search results that include semantic matches related to intent.

Key benefits

• Synonyms and related concepts: NLS offers results that the AI interprets as a match. Such as a search for Bicycle yields items tagged specifically with Bicycle and potentially items tagged with bike or assets with cycling in metadata fields.
• Misspelling correction: NLS recognizes and corrects common misspellings to provide search results based on the interpretation of the search. Such as, a search for Photographt yields results for assets tagged with Photography.
• Long-tail keyword search: Users enter longer, more complicated search terms and obtain results. A query such as 2026 Spring headers photos videos  provides relevant results based on the meaning of those words.
• Multi-language support: NLS understands and interprets queries in different languages. For a DAM system with English metadata, a user can search in Spanish to access search results interpreted through English to Spanish translation. Such as, a search for fotografía returns assets tagged with photography .
• Broader or flow searches: Legacy Acquia DAM keyword-based searches operate on an And flow. This means an asset must have all terms to appear. NLS provides semantic results that cast a wider net. For a search such as Spring mountain bike, NLS returns assets that have Spring OR mountain OR bike in their metadata.

How does natural language search work?

Natural Language Search is only available in the New Search Experience. Users must activate New Search through user preferences to access NLS search results.

• Relevance and sorting: Searches default to the Relevance sort option. This option orders search results with semantic results intermixed. The AI model favors exact keyword matches 20% more than semantic search results. Keywords tend to stay at the top of the page.
• Result limits: Semantic results have a cap of 500 assets per search. Keyword search results have no cap.
• Confidence floor: NLS leverages a confidence floor, ensuring that the AI model provides good additional search results and filters out unrelated or erroneous assets. This ensures only helpful, relevant assets are added to your search results.
• Reversion to exact matches: Users can still use search help shortcuts to return to keyword-match-based searches. Use field titles in shortcuts, such as sku:(462947), to obtain exact lexical searches. Users can also still get more exact keyword matches by enclosing search queries in quotations or brackets, such as {Spring Campaign}.

What is semantically searchable?

NLS uses the existing base of asset information. NLS can search across all assets that have searchable information. The information in this search features:

• Default fields: AI Tags, AI Video Transcripts, Tags, Asset Group Names, Collection Names, and Category Titles.
• Custom metadata fields: Autocompleter Fields, Checkbox Groups, Dropdown Lists, Limited Text Fields, Long Text Fields, Palette Fields, Text Areas, and Text Fields.

Activate Natural Language Search

Because of the AI nature of natural language search, NLS is an opt-in feature.

1. Navigate to Admin > Global Settings > Features.

2. Turn the toggle on for Natural Language Search (NLS).

   
   

The DAM system undergoes a background backfill process to index new and existing asset information. This process does not affect site performance. The administrator account that activates the feature receives an email notification when all assets are processed and the feature is ready to use.

Technical details

NLS uses an AI Large Language Model (LLM) called Qwen3 to vectorize asset information across the Acquia Cloud platform. This is an open-source LLM. It does not use customer data for machine learning. For the initial release, NLS is not available in Portals or through API endpoints.

Natural Language Search in DAM