Build focused, useful collections by choosing the right source content, maintaining it over time, and managing how people access and use it.
Use these practices to build and maintain collections in Knowledge Base in Acrobat. They can help you organize source content, make files easier to find, manage access, and give users a clear body of knowledge to work with.
Use authoritative, current source content
Treat the documents in a collection as source material that users can query through AI Assistant.
Prefer content that is:
- Current and approved for use.
- Relevant to the purpose of the collection.
- Complete enough to provide useful context.
- Free of obsolete or unnecessary duplicate versions.
When a source document changes, update the content in the collection so users aren't relying on an older version.
Use descriptive file names
Choose file names that make the document's subject easy to identify.
Search within Collections searches file names only, not the contents of documents. Descriptive names make it easier to locate and manage files as a collection grows.
Avoid generic names such as document.pdf, final.pdf, or updated-version.pdf when the name doesn't identify what the file contains.
Maintain collections over time
Review collections periodically instead of treating ingestion as a one-time task.
Remove content that is no longer useful, update documents that have changed, and look for duplicate or outdated material. Keeping the collection aligned with its intended purpose makes it easier for owners to manage and for users to understand what information it contains.
Share collections deliberately
Share a collection with the people who need access to its content.
Before broadening access, consider the information contained in the collection, who should be able to use it, and where the collection can be accessed. Organizational settings can also affect access through experiences such as chatbots and PDF Spaces.
For configuration and entitlement requirements, see the related access and sharing documentation rather than managing those requirements from the collection itself.
Use chat attachments for conversation-specific content
Use files and images attached to an AI Assistant chat when the content is primarily relevant to that conversation.
When information should become part of a reusable knowledge domain, add and manage it as collection content instead. This keeps temporary conversational context separate from the maintained source material in the collection.
Check citations when the source matters
Use the citations provided with AI Assistant responses to inspect the supporting source material.
Checking the source is especially useful when the answer affects an important decision, when exact wording matters, or when you need additional context beyond the generated response.
Citations help you verify what information supports an answer; they don't replace review of the source document when accuracy is critical.
Plan for integrations and credit consumption
Consider how people will use a collection before expanding its audience or connecting it to other experiences.
Chatbot integrations, document ingestion, API activity, and queries can have different access and credit requirements. Review the applicable configuration and credit-consumption information before large ingestion projects or broader deployments.
This is particularly useful when a collection will support users outside the group that originally created and maintained it.
Review before expanding a collection
Before adding more content or making a collection available to a wider audience, confirm that:
- The collection still has a clear purpose.
- Its source documents are current and relevant.
- File names clearly identify their contents.
- Obsolete and duplicate content has been removed.
- The intended audience should have access to the information.
- Any required integrations and entitlements are configured.
- Expected usage aligns with your organization's credit model.
