Tesseract is for OCR, so if you use regular PDF, it should be good. But you are right and I mentioned the necessity to add OCR to EmbedChain OSC and to my knowledge they are working on it.