When I was young, I played video games (as probably many other kids my age) and I even designed and programmed my own video games in C++ on my old X86 PC.
At that time, we had some competition between friends debating who’s combat character is more powerful. We often joked about “Who can beat a Level 6 fully geared attacker wizard? Two Level 6 attacker wizards!”
I think today’s AIs work quite similarly. Will anyone ever be able to build a more powerful Generative AI than Chat GPT or Palm 2? And even if someone could, these AIs remain extremely powerful and useful, so better to team up rather than compete with them.
At Eccentex, we believe in AI orchestration.
That’s why we built our own AI (called kAI) in a way that can connects, orchestrate, and work seamlessly together with other 3rd party AI services to jointly deliver the best possible customer and agent experiences.
Nice words, but how does it work in practice?
What does ChatGPT need before it can give you the best answer? It needs an input either in the form of a question or in a form of a command (let’s call it “Prompt”) to generate the right answer or recommendation. And the prompt needs to embed some data / facts as starting point.
But what if the Question (the prompt) and the embedded data is dynamic and changing over time? Or what if you have series of questions that all somehow connected through a logical thread? Or what if the question needs some investigation before you can answer it? What if the answer itself is not enough, because the suggested action(s) needs to be executed in real-time?
Let’s take a simple example.
Imagine that you are a customer of a telecommunication company and are complaining about your mobile network speed. In this case you probably send an email to the service desk like this:
“Hey, my mobile network slowed down and it is regardless of the cellular network strength. I can’t use this low-quality service effectively, so I am really frustrated. If you can’t solve it, I will go to your competitor. James”
As this simple example demonstrates your AI can’t properly answer this claim right away because the input (the prompt) is incomplete.
The company has 567 James’ in the customer database, so kAI needs to run a quick query first, based on the given email address to identify the right James. Then additional customer data should be fetched (maybe from multiple systems) to further analyze what would be the root cause of James’ issue.
Based on the CRM system data, kAI knows that it is James Jones from Atlanta who is using his iPhone 14 with an unlimited data plan, so any limitation on the subscription package cannot be a problem.
Another query from the Cell Location Information system shows James in an area where the network coverage is great and his 5G is enabled on his phone, so no network issue can be assumed either.
An additional query from the Network Information System shows that James’ phone (identified by the EMEI number) seems continuously streaming some data even at night.
Based on the collected information, kAI may answer James’ email like this:
“Hi James, we understand your concern and here to help. Before we start the fault clearance process, could you please check if you haven’t installed some application recently that may use your network intensively, because we see a high incoming traffic from your phone almost continuously, even during the night.”
Without the above-mentioned multi-step Hyper Automation scenario your AI might deliver this response to James’ email:
“Hi James, we are here to help, but unfortunately, we were not able to identify you. Could you please tell us who you are?”
And in the first case, James’ answer might be something like this:
“Ah, you are right, I just installed my bicycle tracking application recently and maybe I left it open in a continuous tracking mode. Now I switched it off and the net is fast again. Thank you for your kind support.”
Now compare that to his potential response to the 2nd case scenario:
“You are my telecommunication provider, and you are even not able to identify me?
Interesting, when you need to send me an invoice you always know who I am, right?
I had enough. I will switch to your competitor because your services are unacceptable!”
As we can learn from this simple example your first response is critical and it can make or break your good CX and mutually beneficial relationship with your customer.
To generate this educated response your AI needs data (from multiple systems). When we are talking about Hyper Automation we mean the synchronized cooperation between various systems, AIs and Process Automation components, and the Decisioning Intelligence that can orchestrate this effectively and autonomously.
So, the answer to the ultimate question:
“What can provide better customer experiences than an industry leading AI engine?”
“Multiple AI and Process Automation Engines that are orchestrated to work together seamlessly.”
If you want to learn more about how Eccentex can deploy Hyper Automation AI in your use cases, schedule a consultancy call or a demo with our Sales team here.
You may also register to our upcoming free virtual webinar through this link.
Tibor Vass – CMO of Eccentex