Upload files to create a knowledge base for agents in Copilot Studio.

Note :

1. You can upload files to SharePoint and OneDrive using the option to add unstructured data as a knowledge source.
2. When you upload documents using this method, the documents will be available to your agent at all levels. However, you can upload documents within Generative Answers nodes. This alternative option gives you more control to specify individual Generative Answers nodes that should not use uploaded documents.
3. You can upload files to groups using the File groups option. 
4. Copilot Studio agents require Dataverse searching to utilize this knowledge source. If you are unable to add Dataverse files to the agent, ask your administrator to enable Dataverse searching in your environment. 

When a user asks a question and the agent doesn't have a predefined topic to use, the agent will generate an answer from the documents you've uploaded. The system uses Generative AI to answer the user's question and provide a conversational-style response.

Uploaded documents are securely stored in the Dataverse. The number of documents you can upload is limited by the file storage capacity available for your Dataverse environment.

To upload individual files as knowledge sources, follow these steps:

1. Open agent.
2. Select Add knowledge from the Overview or Knowledge page , or from the Properties of the Generative Answers node.
3. Upload files by dragging and dropping or browsing to the file location.
4. Add a name. By default, the filename will be used.
5. Add a description. The description should be detailed, especially if Generative AI is enabled, as it helps coordinate the generation process.
6. Select Add to agent to complete adding the knowledge source.

Supported document types

1. Word (doc, docx)
2. Excel (xls, xlsx)
3. PowerPoint (ppt, pptx)
4. PDF (pdf)
5. Text files (.txt, .md, .log)
6. HTML (html, htm)
7. CSV (csv)
8. XML
9. OpenDocument (odt, ods, odp)
10. EPUB (epub)
11. Rich Text Format (rtf)
12. Apple iWork (pages, keys, numbers)
13. JSON (json)
14. YAML (yml, yaml)
15. LaTeX (tex)

Supports images with captions.

Images with embedded captions in PDF files are also supported. Captions, also known as alt-text, are textual descriptions of images. For best results, provide detailed information about the embedded image in the alt-text.

Users can ask questions and receive answers based on information contained in annotated images. For example, the image embedded in the PDF file below provides information about the sales funnel and the stages of the sales process. Users of the agent containing this embedded image as a knowledge source can ask questions such as: "What are the stages in the sales funnel?" The agent can then answer based on the information in the annotated image.

Important note :

1. Images, videos, executable files, and audio files cannot be used as uploaded materials.
2. Images are only supported when they are embedded within a PDF file.
3. Embedded images are supported in all languages.
4. Icons are not supported.
5. Files larger than 512 MB are not supported.
6. Encrypted files, such as those with security labels or password protection, are not supported.
7. An agent can include up to 500 files as a knowledge source.