Just for Clarification, Blue Prism is not an OCR solution. There is an OCR facility within Blue Prism for reading text from screen regions, this technique could be used on a PDF if you can guarantee the quality and position of the text within your document. But, as with all OCR software, you need to test to ensure the quality of the OCR output can be guaranteed.
My recommendation if you need to read text from multiple format documents of various quality is to look for a dedicated OCR data capture tool (rather than a basic OCR archiving tool) or to retain some manual staff to capture the data into a structured format that you can then give to Blue Prism to process.
