Custom Data Extraction

Custom Data Extraction for the Fields Your Workflow Actually Needs

Heyyel Data Extractor lets you build a custom data extraction schema for documents that do not fit a standard preset. Choose the fields, data types and extraction instructions your workflow requires, then return those values as structured JSON or CSV.

Create custom fields → Explore field types

Custom fields Field instructions JSON & CSV 77 languages

Custom / Blank

Extraction schema Custom

customer_name Text

reference_number Text

effective_date Date

amount Number

items Array

Hint: Extract only the products listed in the main details table.

+ Add another field

When presets are not enough

Use Custom Data Extraction for Non-Standard Documents

Some workflows already fit a standard document type. For supplier bills, for example, invoice data extraction may already provide the structure you need. For non-standard business documents, custom data extraction lets you define the exact fields and output structure instead of relying on a predefined schema.

Custom data extraction is designed for documents where you already know what information matters, but a predefined schema does not match your workflow.

Internal documents Extract business-specific references, statuses or operational values.

Industry-specific forms Define fields around terminology unique to your sector.

Mixed document layouts Target information even when the visual structure varies.

Non-standard datasets Build the output around the structure your system expects.

Build your schema

Choose the Right Field Types for Your Custom Extraction Schema

Custom extraction is not limited to simple text values. Structure each field according to the kind of information you want Heyyel Data Extractor to return.

TXT

Text

Names, references, descriptions, addresses, codes or other text values.

123

Number

Amounts, quantities, percentages, scores or other numeric values.

DAT

Date

Issue dates, expiration dates, service periods or business-specific dates.

✓/×

Boolean

Return a clear true or false value when your workflow needs a binary answer.

ENUM

Enum

Restrict the result to one of the values that make sense for your process.

[ ]

Array

Extract repeated items such as products, entries, people or multiple references.

{ }

Nested object

Group related values together when one field contains its own internal structure.

OBJ+

Array of objects

Structure repeated records with multiple subfields, such as line items or events.

Field instructions

Tell Heyyel Data Extractor what a field is supposed to mean

A field name can be clear to your team but ambiguous inside a document. Add extraction hints to guide Heyyel Data Extractor toward the information you actually want, especially when you need to extract specific data from a PDF that contains several similar dates, references or amounts.

Field name alone

amount

The document may contain several totals, fees and values. Which amount should be returned?

date

A file may contain issue dates, service dates, deadlines and signatures.

reference

Documents can contain customer, project, order and internal reference numbers.

Field + instruction

amount

“Return the final amount payable after taxes and discounts.”

date

“Extract the date on which the service becomes active.”

reference

“Return the customer reference shown in the document header.”

One method, many documents

Use Custom Data Extraction Across Different Document Types

Instead of forcing a unique document into a generic preset, build an extraction structure around the information your business actually uses.

Inspection Report

site_name Text

inspection_date Date

issues_found Array

passed Boolean

Operational reports

Capture findings, dates, statuses and repeated observations from internal documents.

Application Form

applicant_name Text

application_id Text

requested_amount Number

category Enum

Custom forms

Extract only the fields needed for your internal application or review process.

Technical Record

asset_id Text

measurements Array

status Enum

next_check Date

Industry-specific documents

Turn specialist documents into structured fields without relying on a standard template.

Your schema becomes the output

Extract Custom Structured Data from PDFs, Scans and Images

The field structure you define determines what comes back. Heyyel Data Extractor uses the same AI document data extraction process to analyze the source file and map relevant information to the schema you created.

Custom extraction to JSON

JSON is useful when your custom schema includes different data types, arrays or nested information.

{
  "customer_name": "Acme Services",
  "reference_number": "PRJ-5842",
  "effective_date": "2026-09-01",
  "amount": 12500,
  "approved": true,
  "items": [
    "Implementation",
    "Support"
  ]
}

Custom extraction to CSV

When your custom fields are tabular, export them to CSV for spreadsheet review or another structured data workflow.

Customer	Reference	Date	Amount	Status
Acme Services	PRJ-5842	01/09/2026	€12,500	Approved
Northstar Ltd.	PRJ-5843	03/09/2026	€8,400	Pending
Atlas Group	PRJ-5844	05/09/2026	€15,200	Approved

PDF data extraction Image data extraction PDF to JSON PDF to CSV Document field extraction

What about scanned documents? For scanned PDFs and images, OCR and data extraction play different roles: OCR makes the visible text readable, then Heyyel Data Extractor identifies the values that match the fields in your schema.

Custom extraction workflow

Build a Custom Extraction Workflow Around the Output You Need

Instead of starting from a fixed parser, start from the output your workflow expects.

Once the fields are defined, the Heyyel Data Extractor extraction workflow continues with language configuration, processing, result review and export.

1

Start with Custom / Blank

No preset schema required

↓

2

Define your fields

Name each value you want returned

↓

3

Choose types and instructions

Shape the extraction around your data

↓

4

Extract your custom structure

Return selected data as JSON or CSV

FAQ

Custom data extraction questions

What is custom data extraction? +

Custom data extraction lets you define the exact fields and output structure you want to retrieve from a document instead of relying on a fixed document template.

When should I use Custom / Blank? +

Use Custom / Blank when your document does not fit one of the standard document presets or when you need a field structure that is specific to your own workflow.

Can I define my own fields? +

Yes. You can define fields based on the information you want to retrieve and adjust the structure whenever your requirements change.

What types of fields can I create? +

You can structure extracted information using field types such as text, numbers, dates, booleans, enums, arrays and nested objects depending on your extraction needs.

Can I add instructions to a field? +

Yes. Field instructions or hints can provide additional context about what information a field should retrieve when a field name alone is not specific enough.

Can I use custom extraction with PDFs and images? +

Yes. Heyyel Data Extractor supports PDF, PNG, JPG, WebP and TIFF files for document data extraction.

Can custom extracted data be returned as JSON? +

Yes. JSON is particularly useful for custom schemas that contain mixed field types, arrays or nested structures.

Can custom extracted data be exported to CSV? +

Yes. Custom extracted data can also be exported as CSV when the selected structure is suitable for tabular use.

Build your own extraction

Define the fields. Shape the output. Extract only what matters.

Start from a blank extraction schema and structure your document data around the needs of your workflow.

Start custom data extraction →

Custom fields Field instructions JSON & CSV 77 languages