I’m currently trying to expand on the Semantic Search demo and would like to display the relevant portion of the PDF that aligns with the string that is returned by the AIP logic function?
Hello! Yes this is possible in no-code, we actually have an example of exactly this. Check out the Use LLMs for entity extraction Build with AIP asset. It even highlights the extracted text from the PDF.
That’s a great question! I love seeing the creative ways people are expanding on Workshop functionality.
To address your scenario: I tackled a similar problem by first parsing the relevant PDF output into a new object set through the pipeline builder. Once that was set up, I returned to the workshop display and added two widgets: an object list and an object viewer. Then, I configured the object viewer to respond to clicks by setting the clicked object as the active one and ensuring it always displayed the currently active object. This approach made it easy to view and interact with the output seamlessly. 
For your use case, if you want to display a specific portion of the PDF aligned with the AIP logic’s string response, you might consider dynamically filtering the object list to match the search string and set the filtered object as the active one for viewing. Depending on your setup, you may need to bind the blobster RID or media reference to the active object for rendering the correct PDF snippet. If you’re aiming for a no-code solution, using calculated properties or Workshop bindings might help you achieve this!