Unified Knowledge Considerations

Sources and Connectors

You can sync and import sources of knowledge from these third-party systems with connectors.

• Adobe Experience Platform (Microsoft Word not supported)
• Amazon S3
• Atlassian Confluence Cloud
• Atlassian Confluence Server
• Atlassian Jira Cloud
• Box
• Custom Source
• Google Drive
• Github
• Guru
• Helpjuice
• Madcap Flare
• MadCap IXIA CCMS (IXIASOFT)
• Microsoft Azure (Microsoft Word not supported)
• Microsoft SharePoint Online
• Microsoft SharePoint Server
• MindTouch
• Salesforce orgs (for Knowledge)
• ServiceNow
• Sitemap (for external websites)
• YouTube 
• Zendesk

For details about each connector, see Zoomin Connector Preparation.

Import and Synchronization

• To import and sync a third-party knowledge source to your Salesforce org with a connector, you need the source’s credentials. In some situations, you also need the login or username, password, URL, security or API token, and anything else required by the external source, such as libraries or page numbers.
• Salesforce syncs knowledge content from third-party sources into the Knowledge Article object every 1 to 2 hours and every 12 hours for sitemaps. Timing varies based on the source and the volume of imported content. We sync articles only if they change. If articles are removed from the external source, they’re removed from Salesforce, too.
• Salesforce syncs only third-party sources that are in a published state. Draft and archived states are ignored. All imported content is read-only in Salesforce.
• When Salesforce syncs and updates articles that are already imported, it creates a new article version that’s associated with the original article. The updated content is published in the new article version, and the old article version is deleted. 
• Salesforce converts all content from third-party sources to HTML, which is loaded into a rich text field on Knowledge articles. PDF files are the only exception. They’re shown on the Knowledge Article record page through an iframe, which loads the text with HTML filtered out into the rich text field so it’s available for grounding Einstein for Service features.
• To prevent third-party content from syncing or updating in Salesforce:
  • Relocate articles to a separate folder in the third-party source.
  • Change the articles’ status to Draft in the third-party source.
  • Implement a filter in the third-party source to exclude the articles.

Generative AI and Einstein for Service

• Third-party knowledge synced with Salesforce is considered for grounding Einstein for Service apps. It plays a foundational role in your Salesforce generative AI experience.
• When third-party knowledge is synced with Salesforce, it’s indexed daily at 1.25 million words an hour for grounding with Einstein for Service features. This means it can take a day or two before you see third-party knowledge appear in your Salesforce generative AI experience.
• Third-party knowledge appears in all Salesforce features integrated with Knowledge, including Einstein generative AI features, Einstein Service Replies, Einstein Search Answers, and Einstein Agent.
• Because third-party knowledge sources are synced into the Knowledge Article object, Salesforce generative AI makes no distinction between external knowledge and Salesforce Knowledge articles.
• Einstein grounds generative content on all knowledge the running user can see. To ensure that knowledge from a certain source isn’t used for grounding AI, use sharing rules to restrict access.

Record Access and Visibility (Security)

• Access to third-party knowledge articles in Salesforce is based on user profiles, data categories, and sharing rules.
• You can define sharing rules for third-party knowledge the same way you define them for Salesforce Knowledge articles. Use sharing rules on the Source field to ensure that support reps and self-service users have access only to the knowledge content they’re supposed to see. See Choose the Sharing or Access Model for Lightning Knowledge.

Search

• Unlike federated search in Salesforce, Unified Knowledge stores copies of third-party sources to help with cross-source relevance ranking in search results.
• Salesforce Knowledge and third-party knowledge sources appear the same in search results for reports, lookup fields, global search, Knowledge list views, the Knowledge component, and Experience Cloud sites. You can filter by Publish Status, Publication Date, Data Category, and Source field in search results. For Experience Cloud, we recommend that customers filter by Record Type rather than Source.

Multilingual Support

• If you added a third-party knowledge source before July 1, 2024, you must delete it and add it again to use multilingual support.
• Multilingual support is available for Salesforce end-user languages, which are listed in Supported Languages.
• Salesforce detects languages from text in the third-party knowledge source and maps them to your active languages in Lightning Knowledge. Images and videos aren't supported.
• If you add a third-party knowledge source connector without turning on multilingual support, you can’t turn it on later.
• After you turn on multilingual support for a third-party knowledge source, you can’t turn it off, but you can remove languages. When you remove a language, any synced articles in that language are deleted from Salesforce.
• If the language of a synced third-party knowledge article changes at its source, it doesn’t update in Salesforce.
• Only articles in selected languages sync from third-party sources; all others are excluded.

Data 360 Integration

• When Unified Knowledge creates an article, it generates a Content Document record. This record stores the article's rich text content, which is referred to as a file. These files are used by Data 360 to ground generative AI features for Einstein for Service. See ContentDocument in the Salesforce Platform Object Reference.
• Unified Knowledge can create a maximum of 200,000 articles each day. See Content Version publication limits.
• You can access Content Document files from the Files related list on Unified Knowledge articles, but they’re not visible on the Files tab.
• We recommend that you don’t delete any Content Document files associated with Unified Knowledge articles, as the articles won’t sync or render correctly.
• If you delete a Unified Knowledge article from the user interface, you also delete its Content Document file and remove it from any linked records, such as cases.