Digitalization is making business processes increasingly dynamic and complex. To be able to maintain an overview and learn from their own corporate data, organizations must embrace complexity instead of trying to reduce it at all costs.
Let this sink in for a second: Just 38 percent of companies are making data-driven decisions to drive their business success. Yet in the foreseeable future hardly any company will celebrate business success without efficiently juggling huge amounts of data. Digitalization is making business processes increasingly dynamic and complex. To be able to maintain an overview and learn from their own corporate data, organizations must embrace complexity instead of trying to reduce it at all costs.
The prerequisite for this is a peaceful partnership between humans and machines. Whilst artificial intelligence takes over repetitive processes and recognizes complex interrelationships, human intellect and creativity do the actual work of interpretation. In this way marketing departments, for example, can keep an eye on which online campaigns sustainably convert interested prospects into paying customers and logistics companies can make reliable statements concerning delivery rates or CO2 consumption on the basis of data-supported predictions. Most managers have long since recognized this potential but many organizations lack data science talent, data strategies or the right tools to implement them.
This is where Fabian Rang and Thorsten Heilig come in: the founders of paretos combine what urgently belongs together in the business world – math genius and entrepreneurial spirit. Rang, the machine learning expert, develops complex optimization algorithms whilst digital entrepreneur, Heilig, knows how to make use of them for business. Their software-as-a-service platform enables companies to use decision augmentation (i.e. AI-powered recommendations for decision-making) to derive optimization measures and future scenarios from existing data. The important feature of this, thanks to its ease of use, is that it makes data analysis accessible even to inexperienced users. In this way, paretos combines the potential of machine learning models with the expertise of business decision-makers and at the same time facilitates collaborative data use across all departmental boundaries.
The idea for paretos was born during a visit to a restaurant. A mutual friend introduced Fabian Rang and Thorsten Heilig with the immortal words: “I think you’ll get along perfectly.” After all, they both share a love of complexity – an inclination that otherwise stimulates only a few friends to engage in lively discussions. Rang talked about the methodical optimization procedures he developed for the rollout of the BMW i3 and the drive system of electrically powered trains with M.I.T, among other things; Heilig – then COO of the Daimler subsidiary moovel/REACH NOW – spoke of the challenges of leading companies into the digital future. “Companies need to engage with the complex world of digitalization in order to master it,” says Thorsten Heilig. “Fabian’s data science methods were, therefore, the solution I was always looking for myself.” A concrete use case, in turn, was what Fabian Rang was looking for to further develop his theoretical algorithm: “It was always clear to me that if my approaches and methods can be automated then a much broader audience can benefit from them.”
The idea of using AI systems to derive optimization measures and future scenarios from existing data is nothing new. But with its multi-dimensional self-learning approach, paretos is setting completely new standards. Because, unlike conventional algorithms that can only evaluate one specific problem, paretos’ “Socrates” algorithm can be applied to many different scenarios and analyzes different factors and variables simultaneously. This allows logistics companies, for example, to evaluate how CO2 consumption, delivery speed and costs should be balanced to be able to increase profit margins. To be able to do this paretos takes on the task of combining all of the dynamic factors that make many digital organizational processes so complex today.
Based on customer data, the software generates a series of optimized forecasts and decision recommendations (scenarios) in the background that all employees can then use to draw the most suitable conclusions for their respective departments. Compared with other AI applications the paretos success rate for optimal choices is a factor of 20-30 higher. In this way companies can achieve around four to five times more efficient results and also, for the first time ever, have the possibility to dynamically and automatically generate new optimal solutions time and again based on their own data.
The precise allocation is made possible by an intermediary “picker” layer that selects a large number of existing models and optimizations. This layer has been trained for tens of thousands of hours and continues to evolve (reinforcement learning). This unique automation allows any operator to use the software for many different applications and makes it the perfect solution for a wide range of companies at low cost.
On average, algorithms developed for individual projects are up to five times as expensive as the licensing fee for paretos. This is because, from a purely mathematical point of view, most business scenarios can be abstracted and categorized very easily. Based on the categorizations and the resulting decisions, AI gradually learns which optimization model is best suited for which use cases and delivers results ever faster, more accurately and more reliably.
For companies, this AI-supported learning process is already critical to their success today. As they look to the future it will be absolutely essential that they outperform the competition. In other words as digitization makes certain processes ever faster and more complex, it is essential that a company’s employees engage with the software as a matter of course which has the added benefit of freeing up heads and hands for creative ideas and innovative thinking.
CTO Fabian Rang already had a permanent place at Daimler after completing his mechanical engineering studies in Karlsruhe. After three years he decided that he wanted to “finally do something useful” and swapped his secure job at the Stuttgart car plant to attain himself a doctorate in systems engineering and mathematics – in line with his father and grandfather (one a physicist, the other a mathematician). His research on “Efficient Multicriteria Optimization for Expensive High-Dimensional Blackbox Problems” already hints at Rang’s fascination with the possibilities of mathematical methods for dealing with complex data structures. He relocated to Lisbon and quickly learned how to write software so he could put his methodological knowledge into practice. He worked on various projects that helped him to perfect his optimization methods and then joined forces with Thorsten Heilig to found paretos.
CEO Thorsten Heilig also has a fondness for digital technology but his approach is more on the business side: How can technological approaches solve complex and real business problems and how can they be scaled? In this context, Heilig’s fascination is with the strategic examination of the topics of organizational and product growth, agile transformation and management of change. He was a co-founder of various companies as well as a systemic management coach and, most recently, COO at moovel/REACH NOW, a subsidiary of Daimler that works on innovative mobility concepts which has made him passionate about exploring new methods and technologies that can be used by practically anyone. With paretos, he sets out to democratize AI-based technology to enable every organization to make better strategic business decisions.
Would you like to know if paretos is the right solution for you? Feel free to schedule a non-binding consultation appointment.