Anna is not human. But she speaks well on the phone
There are companies where artificial intelligence already calls customers. In others, it replaces dozens of consultants. One thing is certain: employees will need more skills in the future.
When people in Germany talk about artificial intelligence, they often express concern about being left behind. As a business location that can no longer compete with the US or China. Or as individual employees whose skills are hardly needed anymore in a highly automated world. “AI is taking our jobs away,” was a recent headline in the “Frankfurter Allgemeine Sonntagszeitung.” “Up to twenty percent fewer jobs – who already has to fear AI,” writes the German newspaper “Die Welt”. It’s true: it’s unlikely that in five years’ time there will be as many translators, illustrators, lawyers, or accountants as there are today. Sooner or later, technology will change every industry – even skeptics and enthusiasts agree on that. But how?
While there is still debate about whether artificial intelligence destroys jobs or merely changes them, and whether the EU regulates the technology enough or too much, many companies have long since moved on. According to a representative survey by the digital association Bitkom, more than one in three German companies now use AI, significantly more than a year ago, when the figure was just 20 percent.
How are algorithms and agents already changing the German economy today? Where are companies using artificial intelligence, what do they hope to gain from it, and what does this mean for employees? With these questions in mind, ZEIT visited four companies: a start-up, a medium-sized business, a DAX-listed corporation, and a small advertising agency. The examples show how much AI has already become part of everyday working life and who benefits or suffers as a result..
The medium-sized company
You might not expect it here, but the future is also being shaped in Delligsen, Lower Saxony, in an industrial park not far from the B3 highway. This is where Bornemann Gewindetechnik GmbH is located. Moritz von Soden, 49, a bearded Hamburg native who quickly switches to informal address and drinks his coffee from an FC St. Pauli fan mug, took over the company with 60 employees from his father-in-law in 2014.
The company became successful, like many other German SMEs, by optimizing a niche product and supplying it worldwide. The threads manufactured in Delligsen are used in German trains, Norwegian oil rigs, chocolate factories, and airplanes.
They are intended to solve a very specific problem there, the “stick-slip effect”: conventional threads sometimes jerk when the spindle and nut rub against each other. This is annoying because it squeaks, and expensive because it causes significant wear. The threads from Delligsen do not jerk. With this message, two sales representatives have so far called companies that use conventional threads and explained to them what Bornemann’s thread technology can do. Classic cold calling. At some point, Moritz von Soden asked himself: Couldn’t we reach more new customers with artificial intelligence?
You can see how this works on his computer screen. Von Soden opens an application that shows him who has visited the company’s website. On this particular day, one of the visitors had the IP address of the pharmaceutical company Bayer. “This is old technology that we’ve been using for years,” says von Soden, “and it’s fully compliant with data protection regulations.” The new thing is that AI agents are now doing their job. They search for buyers who work at Bayer in the thread technician’s address book and in public databases. If they find what they’re looking for, they send the potential customer an email with more info. And if there’s no reply, an AI calls the customer.
Von Soden has named the voice “Anna.” She is supposed to sound like a “young woman in her mid-thirties” with a “warm, engaging tone”; she is supposed to advise, but not sell anything, and disclose early that she is not human. That is what the AI has been programmed to do. Anna speaks 50 languages and knows all DIN standards for threads, which already puts her ahead of most people. Von Soden can follow live in a table who Anna is currently calling and read afterwards about how it went. The person contacted “seems to be familiar with the stick-slip effect,” according to a note from the AI. If interested, Anna suggests a consultation. At this point, at the latest, a human takes over again.
The system has only been running for a few days, but so far Anna has passed every test. She also reacts very confidently when von Soden enters his phone number into the system that afternoon and pretends to be the grumpy owner of a fishing company (“I don’t really feel like talking sales,” “Do you even know anything about my industry?”). Anna recommends threads that are particularly saltwater-resistant and actually sounds surprisingly warm and engaging. Finally, the two agree on an appointment.
“AI is a total gift when used correctly.”
For Moritz von Soden, artificial intelligence is a kind of doping for his team. It compensates for labor that is lacking because it would be too expensive—but it is not intended to replace anyone who already works for him. However, he also says that the more simple tasks are automated, the more highly qualified his employees need to be.
This applies, for example, to Jona Post, 23, who completed her training as an office management assistant at Bornemann. Post is still at the beginning of her professional life, but many of her tasks are already no longer a challenge for artificial intelligence. The Federal Employment Agency estimates that 75 percent of an office administrator’s tasks can be automated. Post is therefore currently undertaking two further training courses, one in online marketing and one in process management with AI.
Jona Post also helped set up the cold calling agents. Von Soden told her what features he wanted, and Post spoke to the developer, made hundreds of test calls, and gathered feedback from customers. The initial feedback was good, she says. Only sometimes does the AI still take too long to respond. “Of course, we still have a lot of fine-tuning to do,” says von Soden, but he trusts the technology enough to let it run for now. “I’m sure it will be really stable in a few weeks.”
And what about the employees who were previously responsible for cold calling? Von Soden says that his two Sales representatives, Tim and Sergii, should have more time for personal consultations in the future, “jetting around the world and talking to people, drinking beer with them and solving problems, not just writing emails.” There will always be enough to do. “Then people will just do something else.”
[…]