Leveraging AI capabilities can dramatically enhance effort-to-value-ratio of sales order processing
Sales orders come in countless forms. Orders come as email and they need to be processed into ERP system. It takes routine human effort to scan these orders and pick up the essential information for further processing.
Sales order processing can be done in several different ways – the most expensive relies on human labor to extract data from sales orders and enter this data to ERP.
We can automate this process
Though traditional capture solutions can also be quite inefficient. This is due to the large number of unique customers sending in sales orders.
For a large international corporation it is not unheard to have much over 20 000 corporate customers that have multiple locations each. In a worst case scenario there will be 400 000 unique sales order templates that must be implemented by capture solution. If each template takes dozen hours total effort will be over half a million hours which is not a good business case.
These routines can be given for artificial intelligence to be handled – human effort is saved for more meaningful and creative tasks.
Efficient sales order processing is achieved combining IBM Datacap Insight Edition with Elinar.AI. The solution uses powerful cognitive capabilities of Datacap to create page layout and tokenize Sales Orders.
Elinar.AI combines IBM Watson with Elinar proprietary AI for advanced understanding of content centric processes. This will take Enterprise Content Management (ECM) onto the next level by leveraging advanced artificial intelligence and machine learning methodologies.
Training artificial intelligence
Initially artificial intelligence has to be trained for understanding various sales orders. This is done using order and invoice history. Large amount of documents are fed to AI. It takes several weeks of super computer capacity for AI to learn sales order differences. We use Power8 Minsky for this.
When AI is trained about different sales order forms it learns to recognize them. If a company gets new partners – new sales orders forms that is – a few samples would initially be run through AI to see if it performs correct or if we need some further training or data post-processing.
AI enhanced sales order processing
When a new order comes in it is scanned with Datacap optical character recognition. After that the trained AI model is able to point out the correct information from the order.
Datacap is further used for validation and verification of the information. After that it extracts the correct information and enters it to ERP or some other system.
Benefits and challenges summed
- Traditional sales order processing takes up huge amount of time with routine work.
- Large international companies might have 20 000 different customers with various different sales order forms.
- Using traditional capture solutions it might take even hundreds of thousands of hours work to implement all the sales order templates into the system.
- When combining capture with artificial intelligence a lot of work effort will be saved for more meaningful tasks.