Freight Paperwork Automation: Testing AI with V7Go & Airtable

Mark Woolnough -

THE PROBLEM: FREIGHT PAPERWORK IS BROKEN, CAN AI ACTUALLY FIX IT?

Well, maybe it’s not broken, but it’s certainly a manual, tedious process. Every shipment involves numerous documents. Commercial invoices, packing lists, bill of lading, certs etc, and they need to be read, checked, and entered into a system of some sort.

AI promises to change that by automating document processing. Automatically extracting key information from documents and transforming into structured data within a companies chosen system.

How’s does this perform in a real-world scenario?

THE EXPERIMENT: CHOOSING AI TOOLS FOR FREIGHT DOCUMENT PROCESSING

There are several tools available all pitching similar solutions, ranging from polished looking enterprise solutions down to budget friendly options.
I narrowed the choice by trying to find ones with some logistics or freight related use cases. I narrowed it down to three options:

  • Rossum
  • Docsumo
  • V7Go

While I’m sure all 3 would have done the job needed for this experiment, I decided to focus on V7Go for this test. Mainly due to I’d heard good words about from a friend, and also I still had trial access.

We also use Airtable to manage some of our business, and they have launched some AI features, so will also test those alongside.

THE TEST: CAN AI HANDLE FREIGHT PAPERWORK?

The test for both options involved the following:

  • Extract some key details such as VAT Numbers, HS Codes, Values, and Weights that could be useful for automating shipping documents
  • Reduce the need for manual data entry
  • Maintain accuracy across different document formats
V7Go Extracting from documents

What Worked Well

  • Document Upload was seamless, and data extractions were smooth.
    Both V7Go and Airtable utilise AI prompts to identify the documents whereby other options give you the ability to map the documents to the data
  • It was fast, documents were uploaded quickly, and it was seconds to extract the information.
  • Accuracy was pretty decent for basic fields in the main. Though, it could select incorrect information occasionally.
  • Automation potential is definitely achievable.
Airtable extracting documents

What Didn’t Work as Well

  • Due to the variability of the information contained in the documents from different clients, sometimes the target information simply was not there.
  • As above, it occasionally selected incorrect information, i.e. extracting an exporters VAT number when the prompt was for the importers (this was because the importer did not have theirs listed - and I’m sure stronger prompting would resolve this)
  • Human verification is very much still needed, While AI was good most of the time, the fact there are errors means that someone is always going to need to at least cast an eye over it.
  • Setup takes effort. AI tools require proper training and structured prompts, which is where most of the work lies.
Errors in HS code extraction

V7GO VS. AIRTABLE: WHICH ONE IS BETTER?

V7GO is a specialist AI tool designed to streamline extracting information from documents in bulk. I’m keen to continue exploring its potential. It offers more advanced capabilities but requires effort to set up properly.

Airtable we are using day to day, so it was pretty to easy to slot it in to our workflow immediately. Even though it’s nowhere near as advanced as V7Go, this ease of use for us makes it a no-brainer to try to implement.

So to summarise, if you're using Airtable already go with that, if not, then a specialist tool like V7Go would be better.

THE VERDICT: IS AI WORTH IT FOR FREIGHT PAPERWORK?

So, could AI replace humans in processing freight documentation? Not entirely!

But it could save a lot of time on this manual data entry and has the potential to automate other parts of the process, such as customs entries, bill instructions client updates etc.

Who Would Benefit the Most?

AI for document processing is best suited to:

  • Businesses handling large volume of paperwork - The more you process, the more you save.
  • Companies with well-structured documents - stating the obvious, but AI works better the data it receives.
  • Companies looking to automate workflows - Once extracted the structured data can be pushed into no code tools to automate things further. 

THE NEXT STEP: USING AI-EXTRACTED DATA FOR AUTOMATION

Now that I’ve seen how data can be extracted, and it’s potential, the next step is to:

  • Find the best ways to use the structured data. Can we get it streamlining customs instructions , booking confirmations, or quote requests?
  • Improve data input quality – If the documents fed into AI are more consistent, the results should improve.
  • Test automation tools like Zapier and Make – If AI can extract reliable data, the next step is automating where that data goes. 

FINAL THOUGHTS

Anyone else tried using AI for document processing in logistics?
Any success stories?