Who responds faster and better: a flesh-and-blood employee or an artificial agent? Just like the KSF, the CCMA also held a technology party last week that offered plenty of opportunity to discuss the impact of AI technology in customer contact.

 

Two consecutive days full of AI technology for customer contact professionals: first the AI ​​Super Session of the KSF on May 22, 2024, and a day later the Digital Conversation Event of the CCMA. Ziptone was only present at the CCMA event, to which the trade press was also invited for the first time.

Customer contact is everywhere

In previous years, the CCMA event was one of the first parts of the 'battle' between the finalists of the CCMA Manager of the Year election. This has been done away with this year. The four finalists were offered a lunch to meet each other, the jury, the board and then the public.

In this edition of the CCMA Manager of the Year election, several customer/supplier/sponsor relationships are striking. For example, Eneco (finalist) does business with sponsors/suppliers Yource, Concentrix and Coniche and Picnic (finalist) has a relationship with (ex)sponsor/supplier CM.com. But the tension among the four finalists was visibly no less.

Dialogue is central

The dialogue was central for the approximately one hundred guests. In itself that is a good choice: there is a lot of noise about AI, but what do operational customer contact managers actually think of it? During the event, the guests were invited after the plenary keynote Sander Hesselink (Concentrix Catalyst) twice over different parallel sessions; in the first round a theme and in the second round a concrete application of AI technology. That dialogue did indeed start in both rounds. Seamly's session discussed the desirability of using AI: should you be cautious because of the difficult to predict impact or should you experiment explicitly so as not to miss the boat? Opinions naturally varied.

9 minutes for a case

One of the parts of the CCMA event: presenting a use case about a concrete application of AI in 9 minutes; a formula that could be used more often. We highlight three use cases.

click to enlarge

At Budget Home it is recognized that the sentiment about the chatbot is not particularly positive, according to Roeland Berkhout. Extra reason to optimally equip the chatbot. Budget Thuis uses four success factors: good demand recognition, extensive testing before going live, good aftercare after the dialogue and a warm transfer. These principles have led to success, especially on specific topics such as information about and adjustment of the installment amount. The next steps that Budget Thuis wants to take: further automate the dialogue and enable the installment amount - including advice - to be adjusted in a chatbot dialogue. Budget Thuis also wants to further explore the possibilities of GenAI.

Wouter Gielen, product owner at ABN Amro Bank, talked about Advisor Assist, which should make agents' lives easier. ABN AMRO has started using several features. First, automated conversation summaries, which are then transferred to the CRM system after approval by the agent. This solution was introduced in May 2023 and produces 15.000 summaries per week, 90 percent of which are correct the first time. The second feature is a smart search function for the knowledge base, which should improve searching in approximately 3000 articles. The system can handle natural language questions. ABN AMRO Bank is also working on implementing call reasoning: useful for gaining even better insight into the 4 to 5 million conversations per year and thus further increasing customer satisfaction. Gielen's tips: always develop together with users and dare to experiment.

click to enlarge

Sonja Green the digital coach presented KPN. No one likes checking off items on a COPC-based quality form. The classic method of call recording and monitoring also means that only a few percent of the calls are assessed. With the digital coach, the dialogues are recorded and converted into text; The innovation in this story is mainly the extensive dashboard with coaches and employees down to the level of conversation facets (such as empathy, quality of the analysis, consent, making an offer). The dashboard shows per conversation where which facets are being discussed in the conversation and makes analyzes possible: does every employee indeed make an upsell offer in every conversation? The implementation took three years and mainly involved a lot of talking, adjusting and adjusting. 80% of the approximately 2400 employees use it and it is not mandatory. KPN wants to further expand the tool with personal support and microlearning ('skills gym').

In the second session that Ziptone attended, KPN and part of the audience discussed the use of automated coaching in KPN's customer service in more detail. Detailed questions were asked about, among other things, how to create support for such an application among employees and the Works Council and to what extent the AI ​​engines in the background are sensitive enough to recognize subtle signals in customer conversations.

Dialogue is desperately needed

You can consider it unfortunate that KSF and CCMA are organizing an event on AI technology almost simultaneously, but on the other hand the conversations on this subject are desperately needed right now. The tone about AI is also gradually changing in the market.

In March, Will Knight published a piece on Wired with the telling title 'Forget Chatbots. AI Agents Are the Future'.
Two months later, OpenAI released ChatGPT4o, whose demos (yes, they are demos) were stunning. Not only did reasoning go remarkably well, the generation of natural speech had also made a huge leap forward. Once again columnist Knight reached for the pen. He, like many of his colleagues from the tech world, was impressed by ChatGPT4o. But in his second piece, Knight became a bit more concrete. Namely:

“It is nothing new for computers to imitate human social interactions, emotions or humor. We're just not used to them doing well,” Knight said. But what are the implications of deceptively lifelike computer interfaces invading our daily lives?

Hardly any latency: extra opportunities and risks

“It will become much more difficult to determine whether you are speaking to a real person on the phone,” Knight warns in his article. “Businesses will certainly want to use flirting bots to sell their stuff, while politicians will likely see them as a way to influence the masses. Criminals will of course also adapt them to set up new scams.” Actress Scarlett Johansson put her finger on the sore spot and pushed OpenAI back into the box. Because the differences are becoming smaller, also in terms of timing. ChatGPT4o can respond to audio input in just 232 milliseconds, with an average of 320 milliseconds, which is comparable to human response time in a conversation.

Can you touch the customer's heart with AI technology?

The discussion has barely started in our country, but will we soon stipulate, for example, that it must be clear to the customer in advance whether the customer service agent has a heartbeat or a clock speed? Or, from a different perspective, as Hesselink presented to the audience: can you touch the customer's heart with AI technology? And for those who want to know what exactly the added value of a customer contact employee is in this age of automation: more about this on Ziptone later this week. (Ziptone/Erik Bouwer)

Corrections and additions: An earlier version of this article stated that CM.com is a sponsor of CCMA. CM.com is no longer a sponsor of the CCMA, the text has been adjusted.

Featured, Technology ccma, Seamly