Orders and goods inbound in 15 seconds: .one AI is transforming your business

Many times in our daily work, we need to upload a supplier document into the system – whether it’s an invoice, a delivery note, a goods receipt, or a purchase order in PDF format.

Have you ever wondered how long this actually takes? And more importantly, how many errors can result from such a repetitive task?

Let’s find out how our .one platform, enhanced with AI, helps users streamline and simplify the entire process!

Once, there was OCR

Until now, document processing has relied on a product known as OCR – short for “Optical Character Recognition” – which, through the use of predefined document templates, has been (and still is) capable of extracting information based on the coordinates provided by the model.

What are the limitations of this technology? The need to define a specific template for each supplier and to create a new one for any new supplier — not to mention the ongoing maintenance required if the document format changes, often resulting in additional costs.

Today, however, we can rely on vertical models known as LLMs — short for “Large Language Models” — specifically trained to extract information without the steps mentioned above, significantly reducing setup costs.

Orders and goods inbound in 15 seconds with .one AI

At Aton, the .one omnichannel platform continues to evolve by fully integrating this technology. As we often explain during meetings with our clients, it is essential to apply Artificial Intelligence where its full potential can truly be leveraged.

From this principle, we developed the engine that processes documents and extracts key information, carefully configured to adapt to bespoke requirements for each supplier.

Why choose .one AI over traditional OCR?

The answer can be summed up in two key points:

• No need to design a template for each supplier
• Specific behaviours can be configured per supplier

In this article, I’d also like to share some suggestions for improvement, based on questions that have emerged during conversations with our clients — questions we’ve already been able to answer thanks to real-life applications of Artificial Intelligence.

• can I extract specific information from the item description in a delivery note?
• how can I retrieve the same piece of information if it’s presented differently by each supplier?

If you’ve found yourself asking these same questions and feel the need to optimise this process with fast, effective solutions, I’d be happy to let you test this technology directly on your own documents!