Single "Item / Function" column. No element columns. LLM used for (2 calls per row — no batch to avoid timeouts): Call 1: Parse focus / lower / higher element names from row text, then match each to ...
Interactive demo for the ai-form-response-extractor library — extract structured data from scanned/photographed paper forms and digital PDFs using multimodal LLMs. Some PDF workflows may require image ...