Skip to content

Work and systems

For carriers and forwarderstransport i spedycja

A wrong photo of the bill of lading is returned to the driver immediately.

The driver sends a photo of the CMR or delivery confirmation. The system first checks whether they can be read. A poor photo comes back asking for a new one. The system reads the data and the operator checks it.

Tested on sample data, without customer data.

test photos stopped as unreadable before someone started reading them
5 of 7
photo features checked before reading
5
for reading one document by AI
USD 0.0076

Syntalith test results on prepared photos of one CMR pattern. Last checked: October 22, 2026.

The bad photo only came to light during settlement.

The operator opened the photo and only then saw that it was dark or cut off. The request to the driver was delayed, but the invoice was waiting.

  1. 01

    Clear photo

    Brightness, sharpness and frame checked immediately.

  2. 02

    Quick request

    The driver knows what to correct before driving away.

  3. 03

    Complete

    It is known which document is missing for settlement.

From photo to settlement in four steps.

  1. 01

    The driver sends a photo

    CMR or delivery confirmation, straight from your phone.

  2. 02

    Photo check

    The system checks brightness, contrast, sharpness, overexposure and frame.

  3. 03

    Data reading

    Poor photo goes back to the driver. The system reads the fields correctly.

  4. 04

    Operator checks

    Each box is marked in the photo. One look is enough.

What the system does and what is left to the operator.

Makes a system

  • Rejects illegible photos with reason
  • Reads fields from a readable document
  • Shows which document is missing

The operator decides

  • Is the reading correct?
  • Whether the document is submitted for settlement
  • Issuing an invoice

Statistics: photo inspection of transport documents in 2026.

According to the Syntalith (2026) test, the system stopped 5 out of 7 photos as unreadable before someone started transcribing them.

Reading one document by AI cost USD 0.0076 in the test.

All test images are from one CMR pattern. We check the quality threshold on the client's photos before implementation.

Scenario: 900 documents per month

Bad photos
approximately 12%, i.e. 108 photos, are returned to the driver immediately
Reading by AI
900 × 0.0076 USD = 6.86 USD, approx. PLN 27
He stays with people
reading check and settlement

Estimate. Assumptions: 900 documents per month and 12% of photos to be retaken are sample numbers. NBP exchange rate from October 21, 2026: PLN 3.8889 per USD 1.

What stays with the operator.

Checking the reading
The operator approves the data before it is sent for settlement.
Invoice
The system does not issue an invoice and does not confirm delivery.
Open thresholds
The team sees and approves when a photo is too weak.

This is what an operator's job looks like.

Screens from the test on sample data.

  1. Sprawdź odczytane pola
    01Sprawdź odczytane polaPola dokumentu są sprawdzone na zdjęciu.
  2. Zatrzymaj nieczytelne zdjęcie
    02Zatrzymaj nieczytelne zdjęcieKierowca wie, jak poprawić zdjęcie.
  3. Sprawdź komplety do rozliczenia
    03Sprawdź komplety do rozliczeniaWiadomo, która przesyłka blokuje fakturę.

Are your invoices also waiting for documents from the route?

Describe your process to us. We will show the prototype on your process and your data. The analysis and prototype cost €0, and you know the fixed implementation price before the contract.

What our OpenAI and Anthropic partnerships mean for you.

OpenAI makes ChatGPT and Anthropic makes Claude. We belong to the partnership programs of both companies, so we work with their tools every day and learn from the creators themselves.

OpenAI Select PartnerClaude Partner Network
01We will select a tool for the task
We implement both ChatGPT and Claude. We check both on your documents, and you choose the one that makes fewer corrections and saves more time.
02Knowledge straight from the creators
We learn from official OpenAI and Anthropic courses and materials. Your team learns about the features and settings that work today.
03Licenses without our margin
When we roll out ChatGPT, Claude or Copilot, you buy the licences directly from OpenAI, Anthropic or Microsoft and they invoice you. We add nothing to the licence.
04Certified engineers
On the team, we have an AWS certification in machine learning and an Anthropic certification in systems architecture on Claude.

Participation in programs does not mean that OpenAI or Anthropic endorses our services.