How does the relationship between EDI and AI work?

Narrow AI is the term used to describe artificial intelligence focused on handling a specific or limited task. It is an approach steadily finding its way into the automation of order and invoice processing alongside EDI.

AI and EDI can both bring greater automation into the sharing of essential documentation among partners in a supply chain.

A prime example of where narrow AI can support EDI is in increasing the efficiency of the accurate flow of trade documents, such as orders and invoices, as businesses move away from a paper-based environment to fully digital.


eDI Instant in minutes

As simple as 1,2,3. Buy EDI online in an instant and connect with your trading partner today

BUY ONLINE NOW

Moving on from manual

Documents sent via PDFs, email, or post are likely to have been keyed in manually and need to be extracted manually too.

There will invariably be ‘typos’ that make product codes or postal addresses ambiguous, something that needs the eyes of an administrator to resolve. For example, the letter ‘O’ can be confused with ‘0’ for zero, or the letter ‘S’ can be mistaken for the number ‘5’.

Manual checking and correction may be feasible where the distribution network or customer base is small, but not in any sizeable business operation with multiple transactions and invoices where it will quickly become resource-heavy and costly.

At the same time, many EDI solutions can struggle to adapt ‘intelligently’ to understand and correct such typos or straighten out other message ambiguities.

AI can support EDI in this scenario by acting as a functional layer to speed up and facilitate the necessary processing and apply standardisation to the order information.

Network management

Here at digital supply chain technology provider Transalis, we already work with clients who benefit from the relationship between EDI and AI in this way.

Among them is the domestic and industrial appliance manufacturer SMEG (UK) whose national distribution network generates hundreds of orders each week using many different systems and protocols.

Narrow AI technology complements the company’s EDI solution, ‘teaching’ the system what it needs to fix including typos involving item numbers and characters at point of entry.

The solution for SMEG essentially ‘feeds’ the system with knowledge of how the particular issue was solved before. This enables the error to be auto-corrected if it happens again. The same approach to reuse and validation is also automatically applied to new customers coming on board.

SMEG (UK) combines EDI with AI

Confidence in order processing

If you’ve been mandated by your retail partner and need to get up to speed quickly, be reassured that help is on hand.

Cerie Paton, Head of Business Systems at SMEG (UK), said the Transalis solution has significantly reduced inbound order error rates from around 15% to well below half a per cent: “After we explained our issues and concerns, Transalis built, tested and put live an API within ten days or so. They worked with us to understand where any issues were coming from, why any data was failing, and how mistakes could be resolved. They have the development and technical support skills to overcome data accuracy issues.”

Cerie added: “Their solution has given us complete confidence that order processing is accurate and essentially error-free. The two main benefits are that we can now easily integrate orders from different customers who are not using structured digital data or EDI, and we have information from retrieved text files that is guaranteed to be 100% correct.”

Benefits of combining two technologies

While the pair have different origins and come to document processing from different angles, the experience of SMEG (UK) shows that EDI and AI can combine to benefit companies and improve supply chain management and efficiency.

The key reasons for this happening now are advances in computer processing capacity, the continual refining of algorithms and ever greater access to data volumes.

In creating a ‘neural network’ to capture structured data from orders and invoices such as reference numbers, delivery dates, item descriptions, quantities and prices, AI paves the way to apply learnings to future processing. And by memorising user actions in response to different data fields, it can increase automation even when presented with unstructured data.

We already see how AI supports critical applications in everything from autonomous vehicles to medical research. The relationship between EDI and AI looks set to bring ever-greater benefits in the automation of document processing too.