Notino’s Jan Hanuš: digital twins will attend meetings for you

Jan Hanuš, who heads the Execution Excellence department at Notino, wants to cut the time it takes to make key decisions from six weeks to a few days. To do so, the online retailer has already deployed hundreds of AI agents and is now testing “digital twins” that will prepare decisions before a meeting even begins. He spoke about it at the Forbes Next Big Thing technology festival.

Hanuš spent a quarter of a century in company management, often at firms that were not in the best shape. For the past year and a half he has thrown himself into artificial intelligence, and his task at Notino sounds simple: speed the company up wherever it makes sense.

Notino overslept the start of AI and shifted into second gear

“The company slept for two years, and at the beginning of this year we decided we had to do something about it. The owner ordered us to shift into second gear,” Hanuš says. According to him, within half a year Notino became one of the leaders in Czechia in automating processes with AI.

Today the company runs hundreds of agents, and dozens of processes have already been mapped with the help of artificial intelligence. The effect is worth tens of millions of crowns. Hanuš sees the greatest value where part of a process can be sped up, scaled or automated.

At the same time, he rejects the idea of a company run mainly by algorithms. A well-trained model can make an average or above-average decision, but an outstanding one still relies on experience, judgment, empathy and a person’s willingness to take a risk. “If we were all, as they say, AI first and had similar models and agents, Notino would be half the size in five years. Companies are still made first and foremost by people,” he says.

From one day to six weeks

More than a thousand people work at Notino’s headquarters, and hundreds of meetings take place every day. In such an organisation, Hanuš says, nobody can any longer see what was decided during the day, whom it affects and where the next step awaits whom.

The result is lost time that has nothing to do with how complex the decision itself is. If all the managers needed sat in one room for two hours, the matter would be closed the same day. In reality, one lacks the materials, another is out of the office, a third needs to bring in one more colleague, and a fourth proposes another meeting at the end of the day. “A decision that could be made in a day or two sometimes drags on for six weeks. Just because people cannot coordinate,” Hanuš explains.

He considers reducing this delay one of the main opportunities of AI in large companies. In parallel, Notino is changing the way decisions are made: as much authority as possible is being pushed down the hierarchy to those who really understand the issue. Previously, too much of the agenda reached the board of directors, but as the company grew, that model became a brake. Artificial intelligence is meant to keep management informed, even if the decisions themselves are made several levels below.

AI training should become stricter

Notino has tested in practice what a purely voluntary approach leads to. Only some employees joined the AI education programme, and participation in the follow-up workshops was lower than the company expected. Some colleagues put AI to maximum use in their work, others did not join at all, and the teams diverged sharply in how fast they adopted the technology. “This voluntariness did not do us much good,” Hanuš sums up.

That is why Notino is considering a more consistent, mandatory approach to training. For different positions, the company wants to gradually define the required level of knowledge and build a system that makes it possible to measure and develop these skills. In Hanuš’s view, in a few years knowledge of AI in a number of professions will be as natural as working in Excel is today.

Digital twins arrive at Notino

Training people alone is not enough for Hanuš. He sees another problem in how quickly employees pass information to each other and coordinate work across the company. The idea of digital twins grew out of his experiments with his own AI tools: a twin is supposed to prepare a decision in advance, so that a meeting does not turn into a long collection of missing data.