PDF files index integrated KB

[Bart]

Hi,

I would like to upload some PDF files to the build-in Knowledge base in JSM Cloud. Are these PDF files indexed so my users can also search trough them when they are creating a Work Item for the Service Desk?

[Christopher Yen]

Hi @Bart ,

So the knowledge base in JSM Cloud is powered by Confluence. If the PDFs are attached I don't think they would be indexed, if they are using the PDF Macros that embeds the PDF it may index the content but the best practice would be to extract and format the data in Confluence natively which should present the data in a better format for your customers. A middle ground may be to at least summarize the KA content into the Confluence page and users can access the PDF if further information is needed.

[Bart]

Hi @Christopher Yen

Thank you for your reply, in this case the PDFs are manuals of products and quite large. So converting them to native Confluence pages is not an option.

I've tried embedding a PDF into the page, but it does not seem to be indexing the PDF file. :-( 

[Lucas Modzelewski _Lumo_]

You can try using ChatGPT to summarize PDFs and suggest tags and questions that the PDF can answer. That would at least help your users find the correct file faster.

I’ve used GPT myself to summarize some training materials - my pages had PDFs for reference, with summaries and code examples, etc. The goal was to get the essence with examples and use cases, while keeping the PDF available for reference if more detail was needed.

I was planning to make an app to "analyze" attached PDFs, with some defined by user system prompts, but with how quickly AI tools are evolving and gaining new capabilities, I figured it would become obsolete too quickly. 😉 But if you need that, let me know!

---

I used something like this:

Instruction:
You are given a PDF exported from a DataCamp course. Your job is to extract the key content (summaries, code examples, and exercises), and reformat it for direct publishing on a Confluence page. Use Atlassian formatting (headings, bullet points, code blocks, etc.).

Prompt Example:

Extract the main sections, summaries, and code examples from this DataCamp PDF.

• Structure the output as a Confluence page using Atlassian wiki markup or standard Confluence formatting (Headings = h1., h2., bullet points, numbered lists, and {code} blocks for code).

• Include a concise summary at the top.

• For each section, add a heading and then the content.

• All code snippets should be in Confluence code blocks.

• Ignore any repetitive DataCamp branding or page numbers.

• If the PDF contains exercises, format them as bullet lists or numbered steps.