As I was packing to go out of town this week I decided to clear up a lingering issue having to do with a $2.17 tube of travel toothpaste I ordered from Amazon. (Yes, as a Prime member it is easier for me to just click “order this” than to put it on the grocery list on the fridge…don’t judge).

Yet here it is a month later and it has never arrived. So I went online to the order, hit “help with this order”, had to enter answer all these security questions and then my chat assistant asked me what the order number was. Really? I had clicked on the order itself to start the chat session.

Yes, this was Amazon.

Was this yet another secret question to make sure I was not trying to scam them out of getting $2.17 refunded to my account for something I might have received and claimed not to? I think not. I think it’s a serious lack of integration problem.

If Amazon, with their two day…or if you live in a major metroplex like I do it could be under an hour… delivery of any type of product on the planet you could ask for, cannot figure out how to integrate their systems so that when a person clicks on an order for help it automatically populates with the order number then who can?

Some conspiracists think that cancer is actually curable, but that the drug companies are having none of that because the treatment dollars are too big. I am not going to get into that debate but I cannot help but think of a parallel between that and the problem of business process integration. Is there a cure for integration between business processes that is secretly being kept under wraps because a multi-billion dollar industry exists to band-aid these processes together?

As it happened, while I was chatting with the Amazon rep about my two dollar toothpaste refund, artificial intelligence and analytics were also on my mind. Then it struck me that perhaps AI could be a solution for the integration problem.

My super personable chat assistant Mekaylor (and yes this was a real person because I asked, either that or it was ELIZA version 2018.7) had to type in a question to me about my order number for inquiry. Yet I can think of several ways that AI/machine learning could have already presented that to Mekaylor: 1) the obvious would actually be integrated systems since I clicked on the order to start the chat, 2) it’s been an outstanding order for over a month, 3) I have clicked on “track status” several times over the last month, 4) I have looked at replacement items…you get the picture.

Think about it. In the time it took Mekaylor to type the question in the chat window asking me my order number in question an AI algorithm could have already put that in front of her.

The problem with APIs is that software constantly changes. The more connected we are the more complex the problem becomes.  It is like trying to keep up with what my 6-old’s favorite character of the week is and trust me, not something easy to do. Keeping up with API changes is something that could be relegated to machines.

APIs are basically a language in which  applications agree on how to exchange information. However if one side changes the agreed upon language – in even the slightest bit – then the integration breaks down and things have to be re-coded, re-tested and re-deployed. But what if a machine quickly learned what the new code word was and automatically adjusted for us?

I realize there are many application integrations in which machine learning may not yet have reached the trust factor. However there are many others, such as my toothpaste issue, in which the efficiency of AI could create a better customer experience as well as improved efficiency for the enterprise. Amazon has no problem using AI to know I have been looking at toothpaste and even suggest an alternate to me so why not use it in their contact center to create a better customer experience?

In the meantime, I will probably go ahead and order the Amazon suggested Crest Incredibles 2 toothpaste while The Incredibles is still the thing of the week.