Extract data from filled forms

A filled form is any completed form that does not have a dedicated extractor of its own. Applications, intake sheets, permit requests, membership forms, questionnaires, and government agency paperwork, anything from a Municipal Business Permit Application to a Vendor Onboarding Questionnaire, all share the same underlying shape: a title and version, a set of labelled fields with the values someone entered, some checkboxes that are ticked or left blank, a few sections that group the questions, and a signature block at the bottom. There are dedicated Talonic tools for the common ones, the I-9 Employment Eligibility Verification form, a benefits enrollment form, or a notarized document, and when a US or EU form matches one of those it should go to the specific tool, because a purpose-built schema captures more than a generic one. The filled-form extractor is the fallback for the long tail of forms that no specific tool covers. What a generic form extractor has to preserve is the structure, not just the text. Each field keeps its label and the value entered, and a value read without its label is useless, so the pairing is the point. Checkboxes are captured as selected or not, because a ticked box is data even when it carries no text. Sections keep their order and their heading, so a response stays attached to the question it answers, and the signature, the signature date, and any witness or notarization are read where present. Because these forms carry no single standard, the extractor works from the layout itself: field labels, checkbox states, section headings, and the completed values as they appear on the page. Talonic reads the filled form and returns the form title and version, the submitter, the checkboxes selected, the field values, and the signature block as typed fields, keeping the sections, the individual form fields, the checkboxes, and any attachments as tables. A completed Municipal Business Permit Application, revision Rev. 2025-03, submitted by Grace Adeyemi to the City of Springfield on 2026-05-06 with the Sole proprietorship and Home-based business boxes ticked, three sections filled, a 75 USD processing fee, and no notarization, returns as structured field-value pairs instead of a flat scan. When a dedicated extractor exists for the form type, Talonic points to it; the filled-form tool is the general fallback, and it structures what the form contains without interpreting it.

What gets extracted from filled forms

How extraction works for filled forms

A filled form can be a scanned paper sheet, a completed PDF form, or an export from a web portal, and because there is no standard behind it, the labels, checkboxes, and sections fall wherever the template put them. Talonic reads the form and maps its parts to the generic-form field set the Field Registry holds, which keeps the form title and version, the submitter block, the signature and authorization, and the completed content as separate fields rather than one text dump. The checkboxes selected and the field values come back as arrays, the form status and the amendment flag are read as their own fields, and the document, signature, effective, and expiration dates parse to ISO 8601. The form sections in order, the individual form fields with their label, type, and value, the checkboxes with their checked state, and any attachments each load as a table, so a section response stays tied to the question it answers. Every value carries a confidence score and a source-region pointer under DIN SPEC 91491, so a reviewer can trace a field value or a ticked box back to its place on the page. When a form matches a dedicated document type, the specific extractor captures more, so Talonic points to it; the filled-form tool structures what the form contains and does not interpret the entries.

Sample extraction

A completed municipal business permit application form

{
  "document_number": "BP-2026-33812",
  "document_date": "2026-05-06",
  "form_title": "Municipal Business Permit Application",
  "form_version": "Rev. 2025-03",
  "submitter.name": "Grace Adeyemi",
  "submitter.contact_email": "grace.adeyemi@example.com",
  "submitter.address": "145 Larkspur Lane, Springfield, IL 62704",
  "signature": "Signed (handwritten)",
  "signature_date": "2026-05-06",
  "form_status": "submitted",
  "is_amendment": false,
  "authorization.notarized": false,
  "checkboxes_selected": [
    "Sole proprietorship",
    "Home-based business"
  ],
  "processing_fee": 75,
  "currency": "USD",
  "form_language": "English",
  "form_sections": [
    {
      "section_id": "S1",
      "section_title": "Applicant Details",
      "section_order": 1,
      "is_completed": true
    },
    {
      "section_id": "S2",
      "section_title": "Business Activity",
      "section_order": 2,
      "is_completed": true
    },
    {
      "section_id": "S3",
      "section_title": "Declarations",
      "section_order": 3,
      "is_completed": true
    }
  ],
  "form_fields": [
    {
      "field_id": "F1",
      "field_name": "Business name",
      "section_id": "S1",
      "field_type": "text",
      "field_value": "Larkspur Home Bakery",
      "is_required": true,
      "is_filled": true
    },
    {
      "field_id": "F2",
      "field_name": "Start date",
      "section_id": "S2",
      "field_type": "date",
      "field_value": "2026-06-01",
      "is_required": true,
      "is_filled": true
    }
  ],
  "checkboxes": [
    {
      "checkbox_id": "C1",
      "checkbox_label": "Sole proprietorship",
      "is_checked": true,
      "section_id": "S1"
    },
    {
      "checkbox_id": "C2",
      "checkbox_label": "Home-based business",
      "is_checked": true,
      "section_id": "S2"
    }
  ]
}

Frequently asked