WythinForge Inbound AI Handler

From delivery note to prefilled fields. Review, then import.

WythinForge Inbound AI Handler: Demonstration interface in Italian, using synthetic data

Watch the demo

WythinForge Inbound AI Handler: demo preview

A 36-second demo. One synthetic PDF, eight checked fields and a CSV for review. Actual local OCR/AI processing. ERP mapping and import must be agreed. Interface in Italian.

Read the demo walkthrough
  1. Before: open the PDF, retype fields into the ERP and check the typing. With Inbound AI Handler: provide the document, let AI extract the data, review and download the CSV.
  2. Upload the synthetic PDF in the current console. The recording shows an actual upload.
  3. OCR, the local qwen2.5:7b model and validation process a configured demo layout. All eight example values match the PDF; this is not an accuracy benchmark.
  4. Download the actual CSV: document 1042002, date 2026-10-29, item 741260 and quantity 34 PZ are among the extracted data. The needs_review status marks the output for human review; ERP import needs agreed field mapping.
  5. Illustrative scenario: 20 documents a day, 220 days a year, 6 minutes before and 2 after yield about 293 hours of theoretical capacity. These are staff work minutes, separate from machine time and subject to measurement in the customer’s process.
  6. Eight fields not retyped mean eight fewer typing steps in this example. This does not mean zero AI errors. Bring a document to define layouts, fields, checks and import.
Watch the WythinForge Inbound AI Handler video

Step inside the process: the problem, the essential actions, the result. With music and captions.

Open or download the video
Read the video transcript

Orders arrive. Data entry changes.

  1. Before: open the PDF, retype fields into the ERP and check the typing. With Inbound AI Handler: provide the document, let AI extract the data, review and download the CSV.
  2. Upload the synthetic PDF in the current console. The recording shows an actual upload.
  3. OCR, the local qwen2.5:7b model and validation process a configured demo layout. All eight example values match the PDF; this is not an accuracy benchmark.
  4. Download the actual CSV: document 1042002, date 2026-10-29, item 741260 and quantity 34 PZ are among the extracted data. The needs_review status marks the output for human review; ERP import needs agreed field mapping.
  5. Illustrative scenario: 20 documents a day, 220 days a year, 6 minutes before and 2 after yield about 293 hours of theoretical capacity. These are staff work minutes, separate from machine time and subject to measurement in the customer’s process.
  6. Eight fields not retyped mean eight fewer typing steps in this example. This does not mean zero AI errors. Bring a document to define layouts, fields, checks and import.

WythinForge. Software forged around your business.

What happens now

The recording produces eight checked values from one synthetic PDF and makes the example CSV available for review.

  • One PDF input

    Process delivery notes using a reviewed supplier layout profile.

  • Eight example fields to review

    Compare the recorded extraction with the PDF, including document number, purchase order, item code and quantity.

  • A CSV for the agreed ERP import

    Download the example output and define field mapping, review and import with your ERP workflow.

Eight populated fields mean eight fewer typing steps in this example. Extracted values require review against the document.

The essential product evidence

Only the most useful screens are shown. Product visuals use synthetic scenarios and may preserve the original Italian interface; customer outcomes remain separate from demo values.

Explore the workflow in three steps

Inbound AI Handler

From delivery note to prefilled fields. Review, then import.

AI proposes the fields; you check them against the document before importing.
Inbound AI Handler: Provide the PDF. Demonstration interface in Italian, using synthetic data
Step 1 / 3Provide the PDF
The synthetic delivery note enters the processing workflow shown. The PDF is available for processing.
What this demonstration proves

The recording shows a synthetic PDF, eight checked values and a CSV for review. ERP field mapping and import method must be agreed. Accuracy on customer documents requires a dedicated evaluation.

How it works in detail

Use the delivery-note PDF as the input

Inbound AI Handler receives a supplier PDF and processes it through a reviewed layout profile. Layouts that are unknown or ambiguous are held for review before the data can move through the agreed import workflow.

Review prefilled fields against their source

The recording shows actual local OCR and AI processing on a synthetic document. All eight expected values were checked against the PDF. Page references and source evidence help the operator review the proposed fields.

Agree the CSV mapping for the ERP

The example CSV remains marked for review. Before operational use, agree field mapping, required checks and the ERP import method. The benefit to assess is staff time spent reviewing proposed values instead of entering every field manually.

Estimate time capacity in your process

What is reducing handoffs worth?

Illustrative scenario, not a measured result. Change the assumptions. The calculation uses staff work minutes and excludes waiting time and parallel tasks.

Editable assumptions

293.3 staff hours / yearPotential capacity released

Theoretical capacity difference under your assumptions.

Assumed uniform monthly accumulation.

4,400 × 4 minutes ÷ 60 = 293.3 staff hours.

Illustrative capacity value: €8,800.

Read the chart values
Cumulative staff-hour difference
MonthStaff hours
124.4
248.9
373.3
497.8
5122.2
6146.7
7171.1
8195.6
9220
10244.4
11268.9
12293.3

This is not a savings guarantee, cash saving or net ROI. Licensing, implementation, training and management costs are excluded. Validate it against measured times and volumes in your process.

Tell us which delivery notes you want to process