For Dr. Tina Klüwer from parlamind, man and machine should master today's flood of communication together.

^{Dr. Tina Klüwer © Parlamind GmbH}

This conclusion was reached as early as 2015 by the expert, who, after studying computer linguistics in Cologne at the Language Technology Lab of the German Research Center for Artificial Intelligence (DFKI), researched chatbots, dialogue systems and text analysis first in Saarbrücken and later in Berlin. Together with her colleague Nuria Bertomeu Castello, Tobias Lehmann and e-commerce expert Christian Wolf, she formed a "powerful team" - parlamind was born. To compare the system with conventional offers such as simple chatbots would be inadequate. Rather, it is a holistic solution for customer service, as she emphasizes. “​Our system understands incoming messages in a very fine granular way and not only sorts them into rough topic blocks, but differentiates according to the specific action wish of the end customer," Klüwer explains the difference, "a keyword search is not sufficient for this. Instead, our machine is based on current state-of-the-art results from research on machine learning and language processing. In addition, it can recognize negative moods and many other information units such as data, locations, customer numbers and more." Whether the contact is made via email, chat or telephone is irrelevant for the artificially intelligent team member - as the system is called on the parlamind website. "Our AI adapts to the company of our customers and can then participate in all main channels of customer service," explains parlamind’s Managing Director. Even more: "Once the customer has trained the system for the email channel, for example, he can reuse the knowledge that the machine already has in other channels.” This enables companies to efficiently process the growing number of communications and save time and money.

Saving time begins with implementation

The advantages are already apparent during implementation, which can be completed in just a few minutes, depending on the customer's system. One decisive reason for this is that parlamind is well prepared for the classic questions of customer service, as Tina Klüwer points out: "The system is delivered "pre-trained" with a set of relevant questions, such as requests for master data changes, and can then immediately understand and process these," reports the expert. "In addition, parlamind already knows some sectors, such as e-commerce, logistics, banks and energy providers, and can be pre-configured for these sectors. This eliminates part of the usual training effort, and the system can be launched comparatively quickly.”

But parlamind’s customers, who currently consist primarily of 60 medium-sized and large companies, not only benefit from the speed and efficiency of implementation for assistance and automation. Through an evaluation function, it also supports the Berlin-based start-up in continuously evaluating the performance of its customer service and thus obtaining better insights, for example with regard to response times. "This reliably identifies optimization potentials and enables rapid follow-up control at management level," says Tina Klüwer. In addition, the evaluation helps to better understand the customers and to respond more individually to their needs. "This enables companies to better organize their customer service and align it closer to the customer, increase efficiency in processing customer inquiries and standardize the quality of the customer experience across channels.” This not only increases customer satisfaction, it also relieves the - human - service team members. "Through the use of our artificial intelligence, the processing of time-consuming and repetitive requests for the agent is reduced to a minimum and the motivation of the service teams is increased," Klüwer sums up, "the time saved can then be invested in the quality of answering more demanding requests and thus increasing customer satisfaction."

When man with the machine...

These "more demanding requests" are a minority. "From a technical point of view, parlamind already enables us to fully automate around 60 per cent of the average volume of enquiries in e-commerce customer service,” the CEO is convinced. But even if, in her opinion, the parlamind systems perform some tasks even better than we humans and the proportion of requests handled by artificial intelligence will increase according to many analysts, she does not want the machine to be understood as competition. "I don't think that man and machine are mutually exclusive, but rather that both sides should control each other and adjust again and again in order to come to the best result together.” This approach is followed by the majority of parlamind’s customers: they rely on "hybrid forms" in which some task areas are processed fully automatically, while others are only supported by the artificially intelligent team member. The customer defines where the threshold lies and how much leeway the machine has in deciding whether to process a request autonomously or to pass it on to a customer service employee. "In this way, we ensure that the AI decision-making process remains comprehensible and controllable also for people at all times," emphasizes Klüwer, and would like to counteract the "hysterical assessment of the effects and possibilities of AI" which she perceives in the current debate.

For Dr. Tina Klüwer, however, this educational work does not end with her role as CEO of parlamind. Since September 2018, she has also been performing this task as a member of the commission of inquiry "Artificial Intelligence - Social Responsibility and Economic, Social and Ecological Potentials" of the German Bundestag: the educational work on AI business models is an important aspect in establishing Germany as an attractive business location for the AI ecosystem. Because: "Only if the AI business model is accepted and viewed positively are customers willing to use the technology and investors prepared to support the company," she would like to give the young, innovative AI companies a voice through her cooperation. The former DFKI researcher knows that scientific know-how also needs to be transferred to an economically viable business model. "Business start-ups are very important, because they are crucial actors for the transfer of innovation from research to commercialisation," says Klüwer, "their flexibility and dynamism enable them to bring innovative products to the market faster than established companies, which have to undergo a complex transformation process." One positive example of this is her adopted home Berlin, which in terms of digitization has a uniquely high density of technology companies and at the same time the largest number of scientific institutions. In addition, the German capital has a good infrastructure and the costs for companies are relatively low by international standards. “Against this background, the aim should be to develop measures that promote further investment in the location in order to achieve synergies between industry and science, which will give future industries such as artificial intelligence further impetus for growth," says Klüwer. With an attractive tax policy, suitable funding and financing possibilities as well as start-up networks she identifies a number of starting points for the whole of Germany. "Bureaucracy and regulatory conditions for start-ups are still slowing down start-ups too much," says the expert. "That's a shame, because once these hurdles have been cleared, there is great potential for innovation and future growth in Germany. Especially with regard to the location conditions for AI companies in Berlin."

“Speaking mind“

The young entrepreneur is certainly not short of ideas. What applies to the German AI landscape in general applies in particular to her own "baby", parlamind. The successful technology start-up is already tinkering with the next milestone and wants to live up even more to its name in the future: parlamind (translated: "speaking mind", "parlare" - Italian for "speaking"; "mind" - English for German "Geist” or “Verstand”) has been working since the end of last year on a voice solution, i.e. on an artificial intelligence that speaks to the customer: "In Germany, around 50 per cent of incoming customer inquiries are still made by telephone, followed by around 40 per cent by e-mail and only 10 per cent via chat and other channels," says Tina Klüwer, "this milestone is a decisive step towards optimizing response times and relieving the workload of customer service agents.“ The system, which was presented to the international public for the first time at CCW 2019, can already conduct its first customer dialogues in German and English, she is pleased to say and adds: "Even after its introduction, talking AI and its further development will remain an essential goal of our omni-channel strategy for us. There is still a lot for us to do here..."