Born in Bergamo, Luca Cortinovis has built much of his career in Logistics Reply, working on the development of technological solutions applied to logistics and management of warehouses. In 2013 he moved to the United States to follow some customers in the automotive sector and since then he closely observed the evolution of the American supply chain and the entry of artificial intelligence into operational processes. In this interview he explains his professional path, explaining how AI can move from the understanding of the context to the recommendation and then to the action and analyzes the most frequent errors of companies that adopt new technologies without having first built an adequate information base. The comparison between the United States and Italy finally leads to a reflection on how warehouses could change over the next five years.
Let’s start from his path. Where does he come from and how did his professional experience in Reply begin?
I was born in Bergamo and started working in Reply almost at the beginning of my career, immediately after the military service, when in Italy it was still mandatory. Since then I started working on technology. Reply was born in Turin in 1996 as a network of companies and the founders had identified from the beginning a tool for growing and bringing innovation. At first the market was linked mainly to the advent of the Internet, automation, automotive and everything that was developing around Milan and Turin, then grew in all directions of digital. Today artificial intelligence is also transforming this kind of solution and Reply is facing it in a favorable way, working directly with technology and together with partners and customers we work with.
His personal path within this development as has evolved?
My path was born in the world of logistics and product solutions. From the beginning I worked on a solution for the management of the warehouse and stayed in Italy for several years until, in 2013, an opportunity for me and for Reply: to move to the United States and begin to follow some customers we had in the automotive sector, bringing in place our solutions. Since then, Reply has grown a lot in the United States, with several offices, including Chicago, Atlanta, Philadelphia, Kansas City, Seattle and Detroit. In this way we continue what was the original mission of the founders: to have inside the network companies strongly specialized, with very specific skills and able to manage the technologies that are proposed on the market, bringing them directly to the customers and collaborating among themselves. Interview with Luca Cortinovis rev
What is more specifically the reality in which it works and how is the logistics sector changing?
I work from the beginning in the reality of Logistics Reply specialized in solutions for logistics. We started from the Warehouse Execution and progressively extended the activities beyond the warehouse: yards, hub, distribution, last mile, up to a level of visibility, to what we can define intelligence in action. The new challenges, even with the introduction of AI, concern above all the orchestration of more actors within the logistics. If we speak, for example, of Warehouse Management, we talk about a solution now known and mature. From the operational point of view, the sector uses these systems practically from the beginning of the introduction of software to support manual activities and traceability. Artificial intelligence is the last transformation.
In concrete, what does artificial intelligence add to systems that have already existed for many years?
In recent years, AI is improving the use of these technologies. He’s not replacing them and he’s not even putting them in quotes, threatening: he’s increasing the ability to do, perform and above all understand. This is the first step and also the way we introduce artificial intelligence within our solutions. The point is to understand the context around the activities we manage, so as to give the operator, supervisor, manager or anyone acting within the network the possibility to interpret that context with a speed that would not be possible simply by questioning a system or looking at information with the human eye. The AI can collect more information and build an extended context, of course with the appropriate guardrail and with well defined boundaries. Once you understand the context you can get to the recommendation: I can suggest which is the best intervention to do. This is where the agent concept comes in.
So the AI agent works with the human operator?
Exactly. We talk about an agent who works with the operator: human supervision, collaboration and recommendation of the best intervention to be done in the face of a certain situation. The next step is action, i.e. to give the artificial intelligence present in our solutions the possibility of acting. But you have to do it securely, and this is fundamental.
In our LEA platform (ndr Logistics Execution Architecture) we formalized a model: “LEA AI Agent Authority Model”. We combine two aspects often treated separately: the maturity in the adoption of artificial intelligence and the level of authority of agents, defining what they can do, when and within which limits.
In a warehouse and in an execution system operations are deterministic, they also concern safety and handling and must bring something from point A to point B. Acting therefore means moving within certain constraints, with a guaranteed authority and managed by those who organized the activity, maintaining human control in the loop: the person, the supervisor, remains within the process and the action takes place through agreed, defined and managed procedures. Interview with Luca Cortinovis rev
Artificial intelligence can enter processes directly. Yet introducing a new technology does not automatically mean having a more efficient supply chain. What are the most common mistakes you see in companies?
It is a very fertile ground of discussion. First of all, we must understand why artificial intelligence is introduced and, of course, one of the main objectives is to reduce costs. In logistics it means generating savings on handling: how many times I touch the goods, like the egg, like I manage it. Recent research indicates that manufacturing and logistics companies are among the ones that have managed to achieve the greatest savings through AI. It means they managed to put artificial intelligence on the ground properly. But where do you get these savings? Seeing more information, automating repetitive processes — I think for example the evaluation of a return — or introducing artificial intelligence in processes not simply to replace them, but to improve them. You do not have to make a copy and paste of the previous process: you have to improve it thanks to technology.
Where does the problem arise?
One of the first problems is the adoption often linked to the hype. Everyone talks about it and everyone wants to introduce artificial intelligence. More than 60 percent of companies are investing in AI in some process or within the enterprise. Very often, however, he does it confusedly. A very high proportion of these projects is still in the proof of concept phase: it is evidence to understand if something can work. What is often missing is the informational maturity of the company behind the introduction of artificial intelligence. In other words, the maturity of the organization is lacking compared to the needs of AI. Interview with Luca Cortinovis rev
Where should a company start before introducing artificial intelligence?
First of all from the data. Many times we start from fragmented systems, from information bases that are not consolidated, from data that cannot be easily aggregated or centralized. Let’s start with information that is incorrectly inserted and then pretend that artificial intelligence gives us perfectly contextual answers to our needs. But this cannot happen if the context behind is not correct. We think of the famous LLM, the linguistic model that allows us to speak with a chatbot: it can give us a perfectly consistent answer from the linguistic point of view, but that answer can not help at all and not produce any value regarding the concrete context in which we are operating.
Can you make an example applied to the management of a warehouse?
I can ask the system why my order has not been taken. The answer could be: you probably don’t have a proper workforce. It’s a plausible answer, but it’s not mapping the information context I really need. What should be able to tell me is, for example: you need an operator at the shelf 12, because that’s where that material must be taken. But to get to that answer you must have the data that informs the system about the presence of the activities and therefore on where the material should be taken. This data must be available in real time and must be linked also with the urgency of the order to be sent, an information that maybe comes from another system. This is the context that allows artificial intelligence to become really useful.
So the company’s preparation comes before technology?
Yes. Before adopting artificial intelligence in its processes, in the world of logistics but also more generally, a company must consolidate its maturity. And it is precisely on these steps that Logistics Reply can intervene. You can start from a very simple level of maturity, almost newborn, and gradually build the different steps until you get to the desired result. The final goal can also be a form of autonomy, but we always talk about controlled and monitored autonomy. Interview with Luca Cortinovis rev
How do you get to this controlled autonomy?
You start to know: from insights, information collection and context building. Then, as the maturity of the organization and systems grows, you can move towards more and more executive activities. Already in the insights phase there may be important returns, but the decisive step comes when automating some processes. This is where the return on investment can become more evident and the ability to make artificial intelligence profitable. That’s the goal you can get.
He has worked for many years both in Italy and the United States. Do you notice differences in the approach to artificial intelligence, in how it is perceived and applied?
With artificial intelligence I worked mainly in the United States, because when this phenomenon began to take on a greater dimension I had already moved here. In general, however, I see a strong curiosity in both contexts. In the United States there is perhaps the perception that AI is more naturally part of the working tissue. There is a strong entrepreneurial orientation, almost from do it yourself, a greater tendency to adopt technologies as quickly as possible: I see that artificial intelligence is useful, I want it and try to use it. If I think of the European and Italian approach, I see more gradual and more reasoned passages in adoption, sometimes even for constraints or limits related to execution. It is a generally more gradual approach.
Does the United States also affect the fact that many of the major companies that are developing AI have been born there?
Surely here there is a strong push on the part of those who are innovating and directly introducing these technologies. Some of the major companies that develop artificial intelligence were born in the United States. In Europe and in Italy, in some respects, we still live in reflex with regard to the production of results, models and technologies of this type. However, beyond the differences in approach and adoption times, I see a great curiosity about AI in both contexts. Interview with Luca Cortinovis rev
If it were to indicate one thing that today many companies still do the old way in the management of the warehouse and that in five years it might seem absurd, which would you choose?
It is a difficult question, because when we enter the reality of the warehouses we realize that, despite years and years of technological innovation, there are still very different situations. Today we have many ways to automate processes, not only through robotics or physical automation, but also simply by automating the execution of tasks. Yet there are still warehouses that do not use a warehouse management system. There are warehouses using Excel sheets to map virtually anything and others that still use huge amounts of paper.
How do you imagine a warehouse in five years?
In five years I would be really surprised if we did not see a strong reduction in the use of traditional devices during operations, regardless of whether to execute them is a person or an automated system. The operator can wear the device or interact directly with the environment without constantly needing a separate device. The automaton, the humanoid or the robotic system will somehow be the device itself and will no longer need the scan or reading as we mean them today. We will move towards a much more visual and much more interactive world. I’d be surprised if this didn’t happen on a large scale. At the same time, however, I expect that even in five years will continue to exist several warehouses still governed by paper.
L’articolo Luca Cortinovis: “Artificial intelligence does not replace logistics, increases the ability to understand and act” comes from IlNewyorkese.
