Survaize: using vision models to generate apps from survey forms

Building apps for electronic data collection of surveys and censuses usually involves a manual process. Subject matter experts create the survey forms on paper (or in word processor) and then someone has to copy each question into a survey authoring tool. This process can take days or weeks and is error prone. Survaize is a proof of concept of using LLM vision models to read a PDF of a survey form and extract the questions, responses and other details and automatically generate an application for electronic data collection that can be run on a mobile device.