Find answers to common questions about accessing, using, and managing Knowledge Base in Acrobat.
This FAQ answers common questions about Knowledge Base in Acrobat, including how the service works, who can use it, how content is managed, and where to find help. Use it as a starting point for understanding the service before moving to more detailed user, administrator, or developer documentation.
General Questions
Knowledge Base is a powerful enterprise-grade Gen AI solution that enables organizations to quickly access and leverage actionable intelligence from PDF-based knowledge collections, supercharge critical document workflows, and accelerate deployment of sophisticated Gen AI solutions.
Knowledge Base supports a broad range of document, data, email, web, and image formats, so teams can work with existing content without converting or restructuring it. Supported content includes:
- Adobe PDF documents (.pdf).
- Microsoft Word, Excel, and PowerPoint files (.doc, .docx, .dot, .dotx, .xls, .xlsm, .xlsx, .ppt, .pptx).
- OpenDocument text and presentations (.odt, .odp).
- Rich Text and plain-text files (.rtf, .txt).
- Data files, including CSV, JSON, and Markdown (.csv, .json, .md).
- Web pages and email messages (.html, .htm, .eml).
- Image files, including JPEG, PNG, TIFF, GIF, WebP, and BMP (.jpg, .jpeg, .png, .tif, .tiff, .gif, .webp, .bmp).
Collections can also include web links (URLs) as sources alongside uploaded files, so they reflect how teams already store and work with information.
You don't need to convert anything yourself — you always upload the original file. As Knowledge Base prepares each file for search and AI insights, some types are handled natively while others are first standardized to PDF automatically.
- Handled natively, without conversion: .pdf, .xls, .xlsm, .xlsx, .csv, .json, .md, .html, .htm, and .eml
- Converted to PDF first: .doc, .docx, .dot, .dotx, .ppt, .pptx, .odt, .odp, .rtf, .txt, and image files (.jpg, .jpeg, .png, .tif, .tiff, .gif, .webp, .bmp)
Whatever the original file type, the result is the same uniform indexing, retrieval, and AI-driven insights.
AI and Performance Questions
Knowledge Base provides advanced AI-powered document insights, including:
- Q&A – Ask questions about document content.
- Summarization – Generate key takeaways from long documents.
- Full-text search – Retrieve relevant document snippets quickly.
- Entity identification and extraction – Automatically recognize and extract key information (API-only).
- Granular attribution – Provides citations for PDF sources within collections.
- Advanced reasoning – Supports complex workflows.
- Web links – Supports internet websites or direct URLs as sources within collections.
- Collections can contain both documents and web links.
- Direct model or internet as sources – Supports using the internet or the LLM itself as a source for answers when content outside of curated collections is needed.
- Model selection – Supports three state-of-the-art OpenAI LLMs: General, Fast Reasoning, and Advanced.
- Format, style, and tone – Supports control of AI output for questions
Knowledge Base is hosted on Amazon Web Services (AWS) and leverages Azure OpenAI for AI processing. AWS is used for core application hosting, data storage, and compute resources, whereas Azure OpenAI provides the large language model capabilities. This ensures scalability, security, and compliance while utilizing enterprise-grade Gen AI models for advanced document intelligence.
Yes.
Ask your question in the language that's natural to you, and by default Knowledge Base answers in that same language — even when the source material it draws from is written in a different language. For example, you can ask in Japanese and get a Japanese answer built from English, German, or mixed-language documents in your collection.
If you'd rather receive the answer in a specific language, just say so in your request (for example, "respond in Spanish") and the answer comes back in that language. A single collection can hold sources in several languages at once, and each source is cited in its original language rather than translated, so you can always check the original wording. The web interface is also available in multiple languages.
The AI Assistant does not directly expose Knowledge Base's Extract Prompt Builder capabilities, but they are available via the API. Users can test document Extract JSON via the "JSON" tab in the AI Assistant on selected documents before using it via the Extract API.
Document and Web Content Handling
Image files are supported today. Knowledge Base accepts standalone image files — including JPEG, PNG, TIFF, GIF, WebP, and BMP — and reads the text they contain so it can be searched and cited like any other document. Images inside documents are handled as part of that document.
Video and audio files are not supported; only text-based content is indexed
Knowledge Base supports web indexing and content ingestion via a Web Indexer desktop application, which can index and ingest authenticated Intranet and Internet websites. When using the Web Indexer, page content is indexed directly into a Knowledge Base Collection.
No, Knowledge Base Web Links don't crawl entire websites. Instead, it retrieves preindexed content from Bing. If you specify a link such as xyz.com, Knowledge Base will fetch the latest indexed content dynamically rather than storing a full copy. Web Links only work with publicly accessible websites, not behind paywalls or authentication
Functionality and Usage
While Knowledge Base is a powerful AI-driven solution, here are some current limitations:
- Data must be ingested into Knowledge Base rather than connecting to live enterprise data sources in real time.
- Limited native integrations (currently Slack and Teams for chatbot integration, API, and SharePoint and Google Drive ingestion).
Refer to the Service Limitations page for a more expansive list.
Shared Credits are an organization-level pool used for metered generative AI activity across eligible Adobe products and services. They provide shared capacity for activities that aren't covered by an individual user's included allowance.
In Knowledge Base, how Shared Credits are used depends on the user's access and the action performed. For example, supported Studio queries can use the user's Fair Use Policy allocation, while activities such as document ingestion and some Slack, Microsoft Teams, or API queries consume Shared Credits.
For the formal definition, licensing model, and eligible Adobe products, see Shared Credit | Product Description
Yes.
Knowledge Base allows users to query multiple collections, which can be done via:
- Web UI – Selecting multiple collections within the interface.
- API Queries – Configuring API calls to retrieve results from multiple collections.
- Advanced Search Filters – Users can specify which collections to pull insights from.
This feature enables users to aggregate insights from different document sets for more
comprehensive analysis.
- Documents per collection: No limit
- Document size limit: up to 1,500 pages per document, and 100MB per file (up to 500MB for PDF files).
- File upload limit: 200 files per upload.
- Number of collections: No limit.
Refer to the Service Limitations page for a more expansive list.
Integrations and Access
Users can interact with Knowledge Base through the following interfaces:
- Web UI – A full-featured experience for managing documents, searching content, and working with AI-powered insights.
- Slack chatbot integration – Ask questions and query documents directly from Slack by launching the Knowledge Base app within Slack channels.
- API – Build custom integrations and automate AI-powered workflows using programmatic access to Knowledge Base capabilities.
Slack and Microsoft Teams are the only direct chat integrations. However, Knowledge Base supports data ingestion through the following methods:
- Web crawling for site-based document sources using the downloadable desktop Indexer application
- Manual SharePoint and Google Drive file selection.
- Static EML email file ingestion.
Knowledge Base requires document ingestion and does not currently support live connections to external data sources.
Yes. JSON is a supported format. You can upload .json files directly and use them for search, Q&A, and citation like any other document
Knowledge Base works by ingesting files into a collection — including PDFs, Office files, HTML, and JSON. Because JSON is already supported, no special arrangement is needed to bring in structured JSON content.
SharePoint and Google Drive are supported for automatic collection sync.
Currently, Knowledge Base collections do not auto-sync with external APIs. However, customers can write applications to ingest, manage or delete content in collections from any source, as needed.
Embedding and User Experience
Yes.
A single Knowledge Base deployment can be accessed via Web UI, API, Slack, and Microsoft Teams. Once data is ingested into a collection, permissions can be managed across different access points without requiring separate deployments.
Analytics and Observability
Yes.
The prerelease version includes basic analytics that show:
- How often are specific documents referenced in AI responses
- Which collections receive the most queries
- Exportable reports (CSV format) for further analysis
Future iterations will include more advanced dashboards for deeper insights.
Admin and Customer Management
- Admins can create and manage collections
- Granular access controls allow different levels of user permissions
- Internal monitoring, alerts, and onboarding tools for solution consultants
Yes, but external customers must be explicitly provisioned. Knowledge Base supports:
- Public-facing domains for external users
- Onboarding and provisioning of external customers
- Customer-facing collection usage metrics
Yes!
You can share collections you create with individual users in your organization, or you can share collections you create with your entire organization if desired. You can also designate other individual users as collection managers with the ability to edit collection contents if you desire.
Getting Started and Support
Users can create a document collection, upload files, and deploy an AI assistant or chatbot in minutes via the Knowledge Base web interface.
Contact your Adobe representative or refer to the Helpx documentation for setup instructions.
