{"id":803,"date":"2023-08-19T09:20:47","date_gmt":"2023-08-19T09:20:47","guid":{"rendered":"http:\/\/18.191.130.61\/?page_id=803"},"modified":"2023-08-19T09:23:55","modified_gmt":"2023-08-19T09:23:55","slug":"guide","status":"publish","type":"page","link":"https:\/\/docudiveai.com\/guide\/","title":{"rendered":"Guide"},"content":{"rendered":"\n

DocuDive AI<\/strong><\/p>\n\n\n\n

Generate an AI on top of any PDF or body of text<\/strong><\/p>\n\n\n\n

Load up any document and DocuDive will generate a custom AI on top of this data so you can quickly find info, summarize content and simplify jargon.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Types of documents<\/strong><\/p>\n\n\n\n

DocuDive is most useful when working with long documents and texts written with inaccessible language. This also accepts images and scanned documents. This usually includes but isn’t limited to research papers, textbooks, annual reports, instruction manuals and legal documents.<\/p>\n\n\n\n

Writing Good Prompts<\/strong><\/p>\n\n\n\n

Be explicit<\/strong><\/p>\n\n\n\n

The trick to interacting with data on DocuDive is to be explicit. If you’re not getting a desired response, simply ask for more of what you want.<\/p>\n\n\n\n

Let’s say you want to summarize a research paper. Instead of saying \u201cSummarize\u201d or \u201cSummarize this research paper\u201d, say something like \u201cExplain this research paper to me in simple terms. What method did the researchers use? What did they find? Why does it matter?\u201d.<\/p>\n\n\n\n

The latter prompt will provide a much better overview of the paper because we’re giving the AI more hints about what we want. We’re using the words \u201cmethod\u201d, \u201cfind\u201d and \u201cmatter\u201d, so now it knows to look in the methods, findings and rationale sections of the paper. See here<\/u><\/a> for more useful prompts.<\/p>\n\n\n\n

Response length<\/strong><\/p>\n\n\n\n

If you’re looking for more or less detail in a response, head into the settings<\/u><\/a>, adjust the max length to suit your needs and then append something like \u201cPlease be as detailed as possible\u201d to the end of the prompt for more detail or \u201cPlease keep your responses brief\u201d for less detail.<\/p>\n\n\n\n

Reading level<\/strong><\/p>\n\n\n\n

Working with a topic that’s new to you? Append \u201cExplain it like I’m five\u201d to your prompt and DocuDive will do exactly that.<\/p>\n\n\n\n

What if you’re familiar with the field but still want to remove some of the complexities in the language? I’ve found that simply increasing the age in \u201cExplain it like I’m [age]\u201d tends to do the trick. Or you can say something like \u201cExplain it at a high-school reading level\u201d or \u201cpost-graduate reading level\u201d.<\/p>\n\n\n\n

If you don’t specify a reading level in the prompt, DocuDive will default to the level used in the source text.<\/p>\n\n\n\n

Break things up<\/strong><\/p>\n\n\n\n

You’ll get much more detailed responses if you break up requests when interacting with DocuDive\u2019s AI. For example, if you’re preparing for an exam and you want to get some practice questions with model answers from your textbook, instead of saying \u201cWrite 20 practice questions with detailed model answers\u201d, first ask for a list of practice questions and then paste them back into the text input one at a time or in groups of two or three. I’m working on a feature to make this process easier in the future, but for now, this is the way to go.<\/p>\n\n\n\n

Useful Prompts<\/strong><\/p>\n\n\n\n

Summarizing and simplifying<\/strong><\/p>\n\n\n\n