The former state-owned company Deutsche Telekom is investing heavily in automated handling of customer contacts. The head of the innovation team is already making arrangements: speaking to a human employee will of course remain possible, but will become 'a option'. 

 

It's nice that Deutsche Telekom pays attention to customers who would like to speak to a human being. But does Deutsche Telekom also take into account that customers may not want to make calls themselves, but use a 'digital companion' for that?

As promising as GenAI may be for the customer contact sector, we are only at the beginning. Yet all signals for even more extensive automation are green. High absenteeism, high turnover, rising wage costs and a tight labor market. From conversations I have recently had with entrepreneurs in customer contact, it is clear that AI is not seen as just a tool ('agent assist'). On the contrary, AI will gradually become the primary delivery mechanism of customer contact. The role of the customer contact employee therefore changes from 'executor' to 'controller' and 'director'. The agent of the future takes action when something threatens to get stuck in a team of chatbots or when a conversation needs to be monitored to ensure it is progressing smoothly. In other words, customer contact employees become process operators.

The customer also automates

Customer contact on the customer side is also changing in character. We are already used to chatbots, self-service environments and interactive web pages. Most of us have no problem talking to a machine – think voice routing or Siri. These applications are still limited: Siri can barely interact with other apps and voice routing cannot connect you to the CEO's mobile number.

The latter is understandable, but the user is not exactly central to these applications. In fact, every company develops its own apps, bots and its own AI assistant – think of Finn, the personal digital assistant from bunq: 'Find out the name of that Indian restaurant where I ate with a friend last year'. Even though Finn and Siri are in the same ecosystem – your smartphone – they don't talk to each other. Let alone that either of them can talk to the reservation app of that Indian restaurant. And provide feedback when confirmation of that reservation has been received.

Rabbit

Accenture expects according to its own 'Technology Vision 2024′ (from January 9 this year) the first chatbots that can work on it will not be available until 2028 task completion – something fundamentally different from conducting a dialogue. The next step, a smartphone that works based on an AI agent instead of standalone apps, is expected in 2031. First, according to Accenture, LLMs must become better at building logical reasoning ('chain of thoughts'). Step two is that LLMs can independently build software solutions for problems (or questions) that they encounter, but cannot yet solve. . Researchers call this 'Large Language Models as Tool Makers' (LATM).

But unfortunately for Accenture: on the same day that the Technoloy Vision was published, Jesse Lyu launched a new device called Rabbit R1 at the CES technology fair in Las Vegas.

Rabbit (“the future of human-machine interface”, photo) is a stand-alone minicomputer based on Large Action Models (LAM). This means that R1 not only understands intentions, but can also independently learn to work with external interfaces and perform actions. In concrete terms: Rabbit R1, as an AI agent, can place an order in your webshop on behalf of his/her owner, have a pizza delivered or have an Uber drive up.

To think about

It is time for contact center managers to start scratching their heads: the 'machine customer' is a fact.
A few things to think about. Is there already a place in both the CRM system and the training program for the machine customer? How does our customer service department deal with out-of-control machine customers – for example in the event of unauthorized use? And if something crazy happens, are we going to call the customer or send a message to the digital rabbit? Should we accept machine customers as customers? Will Rabbit pass our IVR?

And last but not least, have you ordered a Rabbit yet? (Ziptone/Erik Bouwer)

Featured, Opinions, Technology