We are testing a new in-house application that takes any PDF form, and converts it into a web form.
The way this works is pretty high tech and complex.
PDFs are sent to a private bucket (outside of the media library).
A convert job is queued up.
A PDF Worker is scalled up that run s Flock/Ream/Reap skeleton on the PDF, using the same technology we use to scan PDFs for ADA compliance.
AcroForm data is extracted from the PDF using one of our ADA technologies (secret).
Pages get rendered with a special Python library (secret).
5-Page batches are sent to Anthropic Claude servers that are on-demand for EvoGov running on AWS.
A stepper is polled to track the progress (fetch, fillable fields, page images, ai analysis, draft form).
Preview is shown with editable form fields within the EvoCloud CMS for Review
Chance for Review with drag/drop reordering and editing of the web form.
User clicks "Create Form" which sends a JSON package to the EvoCloud import/export server.
Web form is created and marked unpublished.
Once we have this perfected, the process of building web forms from PDFs will be less expensive and much faster.