AI-Driven Document Summarization
Egnyte’s AI-driven document summarization feature synthesizes the content using industry-leading large language models (LLMs). These models will recognize and generate the summary based on the information within the document.
To retrieve the document summary, open any of the supported files for preview and click on the Generate Summary prompt under Ask.
The instruction will be visible in the chat input. Users can make changes, if required.
Users can save the prompt to the Prompt Library, copy the text, or edit the prompt to refine the response generated.
To edit the prompt, click on the Edit icon, modify the prompt, and click on Submit.
The new response will be generated as per the new prompt.
This feature is available on all Platform plans.
AI-Driven Document Q&A
Document Q&A (aka ‘Ask’) feature allows the users to ask questions about a document using a chatbot-style interface. The answers are synthesized using industry-leading large language models (LLMs).
For users that have access to this feature, both the summarization and the Q&A are combined into a single Ask tab. To utilize this feature, click on the Ask tab in the document preview pane.
At this point, users can click on the prompt to generate a summary of the document, or select other prompts. User can also provide audio input or type their questions in the chat box at the bottom.
Users can provide an input and click on the Improve prompt icon to automatically improve and generate a prompt. They can then insert the improved prompt to get the desired response.
Once the user starts interacting on a document, they can also use the suggested prompts to get a response.
Learn more about Prompts Management in Ask.
The users can also use the dropdown at the top to select another class of the document. The prompts will be displayed based on the selection.
Users can provide feedback for every response generated. Click on the New chat button to start a new chat on the same document.
Users can also view the inline citations in the response generated.
Ask on Images
The users can also ask questions on the supported images based on the textual and visual elements in the files. They can use Generate Summary or other prompts to get a response.
The users can also ask various other questions on the image files to get a response or use the suggested prompts once the conversation begins.
Model Selector
Users have the option to choose multiple AI models from the chat input box. The responses will be based as per the AI model selected by the user.
- Model Selector option in Ask is only available for domains with Egnyte Copilot Add-on plan. Please contact the Egnyte account manager or Egnyte Sales team to get started
-
Click on the Model icon in the chat input box.
-
Gemini-2.5 Flash is the model selected by default. Click on any of the other options available to change the model. The model will be changed as per the user’s selection.
Frequently Asked Questions
Is Data Confidentiality Assured while Generating AI-Driven Summaries?
Egnyte remains committed to the confidentiality of the user's data. The inputs, summaries, and
data:
- Are not available to other customers,
- Are not used to improve third-party models, or any third-party products or services,
- Are not stored by third-party models
- Are not accessible by or visible to Egnyte employees
In addition, all generated summaries can only be accessed by users on the domain that have permission to view the original documents.
For more details on Egnyte's Privacy policy, click here.
Is it possible to opt out of these AI features?
Yes. An admin user can turn off these features by navigating to the Settings -> Configuration-> AI page and then setting the Ask AI toggle to Disabled.
Turning the toggle off will turn off both the Document Summarization as well as the Document Q&A features (if available in the plan and on the domain). It is not possible to enable/disable individual features.
What Types of Files Can Be Summarized by the Ask Feature?
The following file categories, extensions, and sizes are supported.
| File Category | File Extension | File Size Limit (MB) |
| Document | "doc", "docx", "docm", "odt", "dotm", "docb" | 64 |
| Spreadsheet | "xls", "xlsx", "xlsm", "ods", "xlt", "xlm", "xltx", "xltm", "xlsb", "xla", "xlam", "xll", "xlw" | 64 |
| Presentation | "ppt", "pptx", "pptm", "odp", "pot", "pps", "potx", "potm", "ppam", "ppsx", "ppsm", "sldx", "sldm" | 64 |
| Plain Text | "htm", "html", "txt", "rtf", "xml", "pages", "numbers", "key", "msg", “eml”, "tsv", "xhtml", "csv", "dot", "wbk", "ps", "epub", "mbox", "wpd", "wp6", "wp7" | 64 |
| GDrive | "gdoc", "gsheet", "gslides" | 64 |
| PDF (text extraction) | "pdf" | 64 |
| Images | "png", "jpg", "jpeg", "tif", "tiff", "bmp", "jpx","gif","webp" | 64 |
| OCR Requeue (ocr) | "pdf" | 64 |
| Construction | "dwg" | 100 |
Construction files (.dwg) are supported only for customers with the Specialized File Handler add-on.
Can Some Documents Be Excluded from Being Summarized?
No. The summarization feature will summarize all files when enabled. Individual files can not be excluded from summarization.
What Types of Files Can Be Summarized and Can Be Used for the Q&A Feature?
Any file where text can be extracted can be summarized, and can also be used for Q&A. This includes text documents, presentations, spreadsheets and PDF files.
Can a user Provide Feedback on these Features?
Yes, the user can provide feedback on the summary and all the responses generated by clicking on the thumbs-up or thumbs-down icons that appear below every response generated by Ask AI.
If the users provide negative feedback on a response, a pop-up appears explaining that Egnyte may review logs to investigate the issue. Users can choose to give consent or opt out. Egnyte will only use logs to address the user issue.
Do the Users Have Any Control Over the Length/Type/Format of the Summary that is Created?
Not at this time. Egnyte is looking at adding additional configuration tools for the users.
Are there any document size limitations for creating summaries?
The first 90,000 characters of a document are used to generate a summary.
Will the Summary always generate the same responses?
Responses are cached and reused. However, the cache can expire and there is a possibility that the generated responses have some variation
Are the generated summaries always accurate?
Like any AI feature, the AI-driven summary is not perfect. It can sometimes generate responses that are incorrect, or misleading. It's important to fact-check information and not rely solely on the responses for critical decisions.
How are the Summaries and Answers Stored, And Who Has Access To Them?
Summaries are cached in memory for performance and are only visible to users who have permission to view the underlying file.
Related Links
AI-Driven Document Summarization in the Android App
AI-Driven Document Summarization & Document Q&A in the iOS App