Vienna In the housing estates of Wiener Wohnen, after all Europe’s largest municipal property management, life takes place to the full. 550,000 inhabitants of the Austrian capital live in one of the apartments. 200 call center employees take care of their worries and needs around the clock. You know the whole range of stressed tenants, for example because the neighbor is constantly barbecuing or the smell of cannabis penetrates the parquet floor.
In February, however, a tenant had a very special problem: the dog howled when you “walked him,” he reported to the call center agent, who recorded the message in writing as usual. Within seconds he had a solution: the tenant should clean the dog’s paws – he was in pain because of the road salt. The employee did not come up with this not exactly obvious idea himself. Artificial intelligence (AI) helped him.
Wiener Wohnen uses software from Deepsearch, also based in Vienna. Natural Language Understanding (NLU) is the name of the process that Deepsearch and other companies use. “Our software analyzes what a text is about in a semantic way,” says founder Roland Fleischhacker.
The keywords dog, go for a walk, howl and the fact that it was February put the system on the right track. And the problem was solved quickly. Fleischhacker is convinced that without NLU a long conversation would have ensued.
Top jobs of the day
Find the best jobs now and
be notified by email.
Unlike industry, the service sector only has a low degree of automation. But that will change in the coming years. Staff is becoming scarce, especially in call centers. If you deal with angry customers every day, you need a thick skin.
Employees change, the know-how of the AI remains
Call centers have fluctuation rates of around 30 percent. The operators train the employees, and as soon as they are in the saddle, many are gone again. Experience goes with them. On the other hand, know-how that is in AI stays in the company. Fleischhacker therefore says: “We relieve the employees of cognitive processes.”
At Wiener Wohnen it used to take around three and a half weeks for a new call center employee to be trained; four days were enough for this, says Stefan Wanner, head of call services at Wiener Wohnen.
>> Read also: Deepfakes: In war we can no longer believe our eyes
The 60-year-old Fleischhacker studied electrical engineering in Vienna and Munich in the 1980s. He then founded Austria’s first SAP consulting firm. Over the past ten years, he has created a so-called knowledge graph with colleagues and his own capital: the basis of deep search and a kind of fishing net that links aspects of knowledge. The company now has 20 employees, but Fleischhacker does not reveal anything about sales.
NLU sounds a bit like science fiction, but has arrived so far in reality that it has created its own industry. Small companies like Deepsearch work with the technology, but also corporations like Microsoft (Luis), IBM (Watson) and Google (Dialogflow).
Fleischhacker says that the tech companies are known worldwide, but they are not really good in the NLU sector. Such statements are hardly verifiable for laypersons: Is it self-promotion of an entrepreneur or a fact? At least the renowned consulting firm Gartner counts Deepsearch among the five leading providers in the world in the NLU sector. The tech giants are not there.
The consulting firm Gartner ranks his company Deepsearch among the top five global providers of natural language understanding.
(Photo: Deepsearch)
The Austrians have already built up a small customer base; in addition to Wiener Wohnen, these include Deutsche Bahn and Die Netzwerkpartner, an organization of 130 German energy suppliers.
Bizarre things sometimes happen at these companies too. What to do when a rail customer complains with the confusing message that he ordered an edamame salad from the on-board bistro but the cheese was missing? In no time at all, the AI guides the call center employee to the menu and the correct section. Edamame is a soybean dish and has nothing to do with Dutch Edam cheese.
Many companies hide the use of AI
As funny as individual examples may seem, companies are always striving to accelerate processes with AI from deep search and other providers. The network partners from Germany, for example, have been using the Austrian software since the beginning of the year to process customer e-mails. They are automatically read and forwarded to the right place.
Employees need a few minutes for this process, the AI completes it in milliseconds. The hit rate of the software is not 100%, but people make mistakes too.
At the same time, companies that use algorithms in customer contact or internally find it difficult to admit this: They fear the accusation that AI is only used to save on staff. A large German pharmaceutical company is a customer of Deepsearch, says founder Fleischhacker, but does not want to make it public.
When AI works, it helps reduce costs. However, the companies vehemently deny that this would involve job cuts. “Rather, we are striving for better advice,” says Stefan Wanner from Wiener Wohnen.
Customers are spoiled these days, explains Claudio Latorre, head of digital products at Die Netzwerkpartner. They always expected an answer as quickly as possible – AI helps companies to meet this requirement. “We want to relieve customer service of routine work.”
A similar development would then take place in the service sector as in the past few decades in industry. But that also means that traditional jobs will definitely disappear, even if companies don’t like to talk about it.
And as in manufacturing, companies will push automation. The call center employees who work with deep search and other providers are still recording customer concerns in writing. At Wiener Wohnen, however, “speech-to-text” is also on the future agenda, says Wanner. What callers say would be automatically converted into text. However, this requires the consent of the customer in order to comply with data protection.
So is it a bit like science fiction? No, Fleischhacker says. One could only talk about this if AI also had a consciousness. “We’re not there yet, and we may never get there.”
More: When the AI tells the baker how many croissants to bake.