Daily Archives: May 28, 2019

AI Will Make Your Talent More Efficient and Effective!

… but, as per our post last week, AI won’t solve your talent problem, so let’s get that out of the way right now!

AI, regardless of whether it’s being sold as assisted intelligence, augmented intelligence, or amplified intelligence, is valuable It’s powerful, and in some domains, essential to the modern day enterprise. In fact, without it, some enterprises won’t survive because the efficiency improvements and related cost reductions it can bring are so significant that competitors who invest will be able to reduce prices and leave their breathren in the dust.

And while it won’t ever replace a workforce, as it will not only allow that workforce to be 2, 3, or 10 times as effective, it will allow you to control the size of your workforce and associated costs where it can be properly applied, leaving more money to hire workers in areas of the business where AI can’t be employed and you are understaffed and/or not operating at full efficiency.

A great example we’re all familiar with is automated invoice matching and processing. A lot of vendors have automated 3-way match and kick out an invoice for manual review when something is off. This allows the department to focus on the 10% to 15% of invoices that have errors instead of spot checking 10% to 15% of invoices they have time to manually review, allowing all errors to be caught and dealt with if the manpower is efficient (and all invoice-based overspend to be prevented, which prevents costly recovery efforts down the road).

But the reality is that only 10% of these invoices have errors that need to be manually addressed. In many cases its missing PO numbers, missing bill to or ship to addresses, unrecognized line items due to OCR errors, etc. which can easily be corrected by an assisted intelligence system which can search the PO database and find out that there’s only one PO for that supplier which is for the same item (quantity, and price) shipped, one address in the vendor master, or a 95% fuzzy match due to a simple OCR error of reading an “O” as a “0” or dropping a character. Why should a human waste time on that? Good AI can handle this, and handle bouncing back an invoice to a vendor when a price is wrong (and allow the vendor to correct or open a dispute), when there are no associated shipping notices or receipts (and tell the vendor the invoice is rejected until such time as one or the other are received), etc. At the end, all the procurement professional or AP clerk is left with are invoices where there are no corresponding contracts, POs, or work orders; invoices from unknown suppliers; invoices with unapproved payment terms (or unknown accounts), and other situations that need to be manually handled. About 1% to 1.5% of invoices, on average for a large organization. For an organization with 1M invoices a year, this is 1,000 – and can be fully dealt with by 2-3 staff members, in comparison to an average organization which would have 20 who couldn’t properly review 20% of the invoices.

There are similar situations where AI can greatly speed up Sourcing Events. For example, new supplier discovery efforts often take 40 to 60 hours of manpower, and a supplier discovery platform like Tealbook can sometimes reduce that to as little as 4 hours and achieve the same result.

There are other examples in sourcing and procurement, but the point is proper AI technology is worth its weight in gold. (But relying on AI to entirely replace part of your workforce will yield an investment less valuable than lumps of coal, which at least you can burn for heat when your business goes bankrupt due to one bad decision on the part of an AI without a complete world-view or real intelligence to process it.)