How indexing works for search

Indexing is the process that makes enterprise content searchable in ZSearch. Once data sources are connected or files are uploaded in ZBrain Builder, ZSearch automatically processes and indexes the content so it can be retrieved quickly, accurately, and securely during search.

Indexing runs in the background and requires no manual intervention once configured.

What happens during indexing

When indexing begins, ZSearch performs a series of coordinated steps for each file or document:

• Content is extracted from supported file types

• Text and structural information are analyzed

• Relevant metadata, such as file name, source, type, and timestamps is captured

• Semantic representations are generated for the full document

• The processed content is stored in the configured search index

This process ensures that both exact keyword matches and contextual, meaning-based queries are supported during search.

Full-document indexing

ZSearch indexes documents at the file level, not as isolated text fragments. Each indexed item represents a complete document, preserving its overall context and meaning.

This approach enables:

• More reliable relevance ranking

• Clear document-level search results

• Accurate grouping of documents into projects

Search results always reference the original source document.

Indexing status and visibility

Indexing is asynchronous and occurs independently for each file. ZBrain Builder displays indexing status so administrators can monitor progress.

Common indexing states include:

• Pending – The file is queued for processing

• In Progress – The file is currently being indexed

• Indexed – The file is fully processed and searchable

• Error – Indexing failed and requires attention

Only files marked as Indexed are available in ZSearch.

Permission-aware indexing

During indexing, ZSearch also captures permission metadata from the source system.

This ensures that:

• Access controls from the original data source are preserved

• Users only see documents they are authorized to access

• Search results and AI-generated answers remain permission-aware

Permissions are enforced at query time, not duplicated or overridden during indexing.

Index freshness and updates

Indexing is not a one-time operation. ZSearch continuously monitors connected data sources for changes.

When new or updated content is detected:

• The affected files are re-indexed automatically

• Search results are updated to reflect the latest content

• No manual reindexing is required

This ensures that ZSearch always operates on current enterprise knowledge.

How indexed content is used in Search

Once indexed, content becomes immediately available for:

• Keyword and natural language search

• Filtering by data source, type, or recency

• Inclusion in collaborative project workspaces

• AI-assisted querying grounded in indexed documents

Indexing is the foundation that enables accurate retrieval, relevance ranking, and secure collaboration in ZSearch.

After understanding how indexing works, the next step is to learn how ZSearch keeps indexed content up to date over time.

The next section covers:

• Syncing and updating a knowledge base