About AI Search Index
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-25-2025 04:22 PM
I am researching AI search indexes.
Is this different from the text indexes that are created for each table?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-25-2025 09:04 PM
Hi @bonsai ,
**AI Search Indexes** and **Text Indexes** in ServiceNow both enhance search functionality but serve different purposes. **Text Indexes** are database-level structures created on specific text fields (e.g., `short_description`) to optimize full-text search queries within a single table. They tokenize text for faster keyword or phrase searches, improving performance for queries like `123TEXTQUERY321`. However, they are limited to specific fields, lack natural language understanding, and cannot search across tables or unstructured data like attachments.
**AI Search Indexes**, on the other hand, are part of ServiceNow’s AI Search functionality, built on **Elasticsearch** and powered by machine learning and natural language processing (NLP). They enable intelligent, context-aware searches across structured and unstructured data, including text fields, related records, and attachments. AI Search understands user intent, synonyms, and context, providing highly relevant results even for ambiguous or natural language queries (e.g., “how to reset my password”). It also supports relevance scoring and cross-table searches, making it ideal for the **Global Search** bar.
The key differences are:
- **Text Indexes** optimize full-text search on specific fields, while **AI Search Indexes** enable intelligent, cross-table searches with NLP.
- **Text Indexes** are manually created and field-specific, whereas **AI Search Indexes** are automatically maintained and span multiple tables and attachments.
- **Text Indexes** are ideal for structured keyword searches, while **AI Search Indexes** excel at handling unstructured data and natural language queries.
Please mark it as "Accept as Solution" and "Helpful" if this solution helped you.
Regards,
Satya
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-25-2025 09:10 PM
Hi @bonsai ,
Index and search record data by table. The ServiceNow Zing text indexing and search engine is the default search engine used to search Now Platform record data. Its engine is created entirely with ServiceNow code and implemented within the relational model.
The Zing text search index is split into multiple database shards for parallel querying. Its search engine logic yields fast results and emphasizes relevancy scoring. Relevancy scores include frequency and proximity weighting. Scores also include weighting for specific indexed fields such as title, short description, and metadata.
Zing provides type-ahead search suggestions and Did you mean: global search suggestions.
Whereas, The AI Search index stores data from Now Platform records or external sources and makes that data available for users to search in multiple applications. Search query features use Now Platform technologies to improve the search user experience.
To help you with your Research, refer to this Doc, which has all types of Search available in the platform along with it's architecture:
