Inside Logistics AI: The Industry of Sleepless Nights.

Cristina

Thank you very much.

Cristina

There is a saying, you know, that when everybody else is sleeping, logistics people are working behind to move the cargo from here to there. So it’s definitely a huge industry. I’ve been part of it since two thousand nine. So a lot of years already, jumping from one to the next: land transport, air freight, sea freight, logistics, customs formalities, port operations, and it gets more and more complex. So that’s what I did in the past years at a freight forwarder. I’ve been part of a forwarder. And this year in FlowX.AI, I came to, let’s say, to apply my expertise to the use cases that we have developed here and that we are still developing in order to gain as much potential customers as possible in the transport and logistics industry.

Cristina

Speed.

Cristina

Speed of documents, speed of processing. Okay, not the speed of trucks maybe or in rail freight, but the speed at which things happen. Speed and volume, let’s say. What what happens at present is that in logistics they are able to process more orders, more volumes, more containers due to the advancement in technology compared to fifteen, seventeen years ago.

Cristina

It is the perfect match and the perfect industry for sure. That’s why I think it was very easy in FlowX.AI to identify a lot of use cases, to cherry pick somehow the topics that we want to dive into because the potential is huge. On the other hand, some solutions were already there or started to appear in the last five to ten years, let’s say. So what we want to bring to the customer is something additional, something they did not try already. If we talk about, I don’t know, truck optimization that’s been going on. But there’s more to be done to that.

Cristina

First thing I noticed when I came here, everybody said, oh, but these client these logistics clients are cancelling a lot of the meetings. Yes, because things always happen. Something is burning somewhere in logistics. So in a bank, you fix a meeting three months before you’ve got a big round table of participants and it happens. In logistics, there is this uncertain element. But looking a bit deeper into the topic, what is different maybe faster possible implementation because the systems they work with are, let’s say, ready to be connected with and extract data. Maybe faster faster decisions from this management layers point of view. You can get a project approved in, I don’t know, six, eight weeks maybe. If you get the right you can get the right signature in this shorter time. What else? I mean, there is sensitive data, but not so sensitive like bank accounts and IDs. We work with prices, we look at the pallets, we maybe look at transport routes. Of course, there is some concern behind the the data protection, but I would say it’s easier to cover in logistics.

Cristina

There is indeed millions, millions of shipments, millions of invoices and corresponding transport documents. So when we say reconciliation in logistics, it may not mean looking at only the order lines that you find in the invoice, but it may mean cross-checking it with a document like the CMR or the delivery note. Or if you want to make sure that what was ordered is invoiced, you go and check it with a PO maybe a purchase order or stuff like that. So it’s a whole set of documents behind in the file that needs to be cross-checked and matched. In order for that invoice to be approved and to get paid.

Cristina

Well, you know, I mean, in a logistics operator, you have the operational team and behind is the admin desk or the finance team who has to move to keep the speed when processing these documents. So they arrive on their desk. Maybe they do not they are not complete in terms of, let’s say, details. Maybe there’s missing information. Maybe there are some conflicts. And this is where FlowX.AI is stepping in in order to cross check all these details, all these requirements in Incoterms. Maybe, I don’t know, there was a discount for that client or there are some contract rates agreed which were not applied at the moment when the invoice was issued. All these sort of things we can do now automatically. Of course, if the agent has been well trained and taught at the beginning, we always tell our customers, give us access to your data, to your rules, to your contracts in order to train our agent to make the best decisions and to make sure that an invoice is not approved on the flow with some maybe mistakes or errors happening there in the back that we didn’t catch.

Cristina

Well, that’s actually one of the good let’s say one of the parts where the agent is very good, maybe identifying the missing data information and putting it into the puzzle. An example. If we have, let’s say, an ocean freight shipment, yeah, coming from Asia to Europe, you have a bunch of costs there in the transport invoice. Some of them are variable from one transport to the next. Some of them are constant, like port charges or maybe maybe this part of inland transport. That piece of cost, you can find it in the system of the provider or the 3PL or the carrier. You can identify it, you can maybe cross check against some rules and you can add it to the let’s say to the invoice or to the calculation if it’s missing. This is the the beautiful part that the information is lying somewhere in the systems. In the TMS or maybe in the customs systems and can be identified and put together in the puzzle.

Cristina

You’re definitely getting it right. So, this info is available and usually it’s not lying in one single department. This is the beauty, that if we are missing information when we reconcile any type of documents, not necessarily invoice, we pick maybe some data from the purchasing department. We look into a client contract for a price or some other accessorials. We look maybe into the fiscal finance department in the previous shipments of that certain customer. So this is how we can identify it. And every logistics operator has their clients, yeah, among these clients. Usually the biggest ones are the ones who are not playing by the rules. There’s some missing information, they send the orders past the cutoff time, and they are these are the clients that expect you to do the extra mile. Like, figure it out for yourself. Yeah, we’ve been working for 10 years. You know what our rates are, or you know how we invoice these accessorials. Go to the contract and look for that. This is kind of how the how the relationship goes. And the clients are starting to have this expectation from a 3PL, from a logistics provider, to figure out the missing parts.

Cristina

I mean, the impact can go, I don’t know, as high as, let’s say, speeding up the process three to four times or cutting this manual process down to, I don’t know, ten percent of the initial time. Some some invoice or some documents that been checked and uploaded into a system in nine or ten minutes can now be can now be processed in seconds, maybe twenty, thirty seconds, something like this. So imagine the magnitude of the saving, of the improvement. And probably in logistics in one, two years from now, the gap between the companies that are doing this and the companies that remained in the old way of working will increase like I don’t know exponentially.

Cristina

Yeah.

Cristina

Yeah, first I mean first we look into the systems they use at the moment. If they have some modern software or pretty capable WMSs or if they collaborate with their clients in already a modern way, it’s a sign they are open to AI and to these kind of technologies. If their standard way of operating is pretty obsolete, let’s say, or they work the same now like they worked 10 years ago. It’s a sign they may not be very open to.

Cristina

Yeah, totally. I mean, they know what it means, they know what are the risks and the limits. They’ve tested it before maybe. We have some of the clients say we tried this solution, I don’t know, automatic dispatch or direct quoting, stuff like that, and we know what are the risks, what are the benefits. It’s a good territory for us to explore further. Or if you go and meet them on site in their premises and you see their desks are full of stacks of paper, that’s another sign there there’s a lot of potential there for automation.

Cristina

Yeah, totally. If you if you get the chance to see their office, you you can at the end of the meeting, you can have an idea about how big are their automation needs and agentic needs.

Cristina

Or exponentially easier.

Cristina

Yeah, the distribution world is, let’s say, an entire chapter in itself.

Cristina

One of the hardest, one of the hardest. Yes. Things are moving very fast because everybody wants the goods on the shelf yesterday. So we’re talking about two flows. Yes. It’s the flow of goods from supplier to the warehouse to the store or to the destination. And there’s the flow of documents. You place an order, you issue a purchase order, you expect the supplier to confirm it or to make some changes, they ship the goods, you receive them in the warehouse. Yes. So there’s a question how fast are the goods moving? How fast are the documents moving through the system? Yeah. Because if if the operator is the delaying this check, for example, you didn’t check today what you will receive in the warehouse, the truck is already in the distribution center, the driver is yelling to be released because he’s got another load planned in the afternoon, and somebody in the warehouse says, okay, let’s offload the goods, put them somewhere there in the reception area, and we’ll figure out later. If it’s complete, if it’s what we ordered, if we are missing one pallet or maybe if we received a a totally different product. So that’s how these discrepancies are happening and somebody has to go back every time and do this check. And what we what FlowX.AI can do is kind of prevent this kind of situation. Because a P it a PO can be matched in seconds. An ASN can be checked again in seconds or in the night when the order was placed. And you have a lot of time until the goods actually reach a warehouse. I mean, in logistics we’re talking about transit times of thirty days plus, if it’s an ocean freight shipment or even in a air freight, it’s like can be twenty four hours. There’s plenty of time to do these checks, but when it’s done manually, some delays can happen. And then, if the goods arrive faster than the documents were validated, that’s where it comes up.

Cristina

How do we solve this? We first we need to connect all these systems. And this is where FlowX.AI is coming with the top layer, let’s say, over your ERP maybe, the WMS, the warehouse management software and maybe the financial software or the supply chain

Cristina

In most of the cases, yes. And this is where is our competitive advantage.

Cristina

It can be. But if it’s assessed technically correct from the very beginning, it’s just a normal integration phase when we sign a new customer. I mean we identify how many systems do you work with, how do these connect to do they offer API or any other type of connectivity, how different is the data from one source to the next? Because in a system maybe you call the truck truck. In another system, you may call it FTL. So FlowX.AI will need to do this match and to understand that truck and FTL are one and the same thing. So I think the secret here and the key is to have this assessment done correctly on time before we go live, it can be done in the implementation phase. In the testing phase, the client gets a testing phase where all these things can be tested, amended, corrected, so that when we go live, things are functioning no matter if we put together like three systems or maybe ten to work together simultaneously.

Cristina

[unclear], yeah, I mean it’s directly reported, let’s say, to the stock of goods or the value of the goods they are moving because any error means, I don’t know, millions of dollars or euros of cargo stuck somewhere on a floor, not visible in any system anymore. Because the supplier took it out of its system already. In your warehouse, it’s not checked in. So it’s cargo somewhere in limbo.

Cristina

Exactly. All of that, it’s money blocked somewhere in the warehouse floor or maybe in a truck waiting in customs. Another example of blockage.

Cristina

A must. A big must. Well, I think it can take them from you know, in logistics we were saying we cannot fill up this warehouse to one hundred and twenty percent capacity, there is a cap. But with this technology basically, I think we can start going over one hundred percent because you optimize the space so well, you optimize the trucks so well and the routes that some competitors might look and say what are they doing and we’re not doing? How come their truck does never have a one pallet empty space? Or how come their warehouse is full no matter if it’s summer time or winter before Black Friday? You can reach these kind of synergies that now, mathematically, they seem impossible.

Cristina

It’s transforming, and I think it will dictate in the future what providers stay in the business and what other providers are going out of business.

Cristina

Totally, yeah. It’s like, you know, if you are a carrier and you have your fleet of fifty trucks, if these don’t move around all the time, they sit idle, they will accumulate costs so much that they will simply take you out of business and then the leasing company is taking back the trucks. It works the same for an air freight operator, for logistics provider, for a I don’t know, customs broker. Any kind of operator because not all the players are so lucky to have like electronics or microchips, right, in the trucks or in the warehouse. Logistics is also moving lower value cargo, lower value and high volume. When you look at construction materials, you look at some pipes, I don’t know, even textiles. This is the type of cargo that cannot have, I don’t know, 15% EBIT margin on top. The margins are much lower, and there will be this segregation in the next years. Who’s staying in business?

Cristina

I mean, I’ve been saying this since 2009, but coming into the tech and AI industry, it’s definitely like I don’t know history being written. And you know, the continents are already at different levels. We have customers in the US. I think they are maybe six to twelve months ahead of the customers in Europe, for example. They have tested already some AI solutions, they have lost their first first money in such projects, they have, let’s say, pioneered some, I don’t know, breakthrough technologies and now they are already getting into their second contracts or already tendering for next year, for twenty twenty seven. In Europe, I think companies are touching base for the first time now.

Cristina

It’s still a tipping point, yeah. Everybody’s curious, everybody’s a bit reluctant. That’s why we I mean we have to be very convincing when we do the demo, when we do the presentation to the customer and we try to to convince them to say yes, yes to one project. Maybe a smaller project or a project that is not presenting too high risk for them, and then they will see that everything goes well, maybe above their expectations, and they go into the second, into the third project. You don’t want to risk like all your business at once. This is how the customers see it, and we adapt to that.

Cristina

I would say it’s a myth. If you ask the experienced broker, of course, no software can make the decision better than they are doing. And there’s a need, you know, for new people, for newcomers in the industry. In every company, you see that. Thirty years plus experienced broker who thinks nobody can match a truck and a load better than he can. But the truth is that AI is simply thinking like, I don’t know, one thousand times faster and checking one thousand bigger quantity of data from the system and maybe checking the market in real time, it’s it’s impossible for the human brain or you cannot do this. You cannot speak on the phone with ten different carriers in the same time to see who has the truck available for your load.

Cristina

Well, the beautiful part about this agent is that is that they learn a lot every time, with every single cycle and flow they learn. For example, let’s say we do some quoting, yeah? The agent is suggesting some selling price that the broker is supposed to send back to the client. We come up with one recommendation or three, the agent comes up with this recommendation and the broker says let’s go for option four because and it should input some reason there in our platform, in the dashboard, ya? Okay, the agent next time he gets the same inquiry from that customer, it will know already that there is this option number four, which can be more profitable and it will go to that. So it’s it’s important to train this agent because it’s not like a one size fits all, right? No solution from customer A and B will fit to customer C. Somebody there with the wisdom, with the tribal knowledge, as we call it, has to put that knowledge to work together with the agent. If not, the agent will bring non-stop the same recommendations every time you run it. But if somebody, the senior people, are training that the newcomers, maybe in the sales team or in the company, will benefit out of that.

Cristina

How we do it in in this testing phase or implementation phase, we ask them to work. Normal day to day operations and we shadow and we see what they do. So we try to copy, copy and paste into the agent all their actions. Let’s say you want, I don’t know, some extra capacity on some some certain routes. We are looking at which carriers is contacting, which carriers is not contacting. We are trying to identify patterns, we can ask questions. Why did you select this option A and not option B? We can look at the statistics accumulated because they usually have years and years of shipments of data from the past and the data can give you a lot. And then we make assumptions based on this data, but we cross check or we clarify and confirm these assumptions with that experienced broker who can give us the how to say this confidence or this accuracy going up from ninety five percent maybe to ninety nine percent or to perfection, close to perfection.

Cristina

Exactly, exactly. And we talk to them about this return on investment, about this volume and about this growth, to make them understand that, okay, you are handling, let’s say, 10,000 shipments a month, the human in the loop doesn’t have to confirm each and every one of those ten thousand shipments, but they will confirm the exceptions. Exceptions are always in the range of five percent to ten percent, let’s say. This is the beauty, that the other ninety percent of the shipments or invoices or transports they go in the automatic flow, and then the human can focus only on the exception. And this is creating huge extra time, let’s say, for them and synergies. Basically, allowing the client to scale up without adding extra workforce.

Cristina

Well, I mean, the impact will be different on the different players. If you are an airline or a shipping line, definitely you are more advanced in this technology and you are already reaping the benefits. If you are, let’s say, a direct carrier, you own a fleet bigger or smaller, or you have some warehouses that are not still optimized or at full capacity, you will see, let’s say, bigger benefits. But the beautiful part, I think, that is this technology applies equally to a small or medium sized player who wants to grow, let’s say, ten times next year, and that makes it possible because it gives them the access to quote ten ten times more to tender or to contract other clients, bigger clients and so on. But the technology is equally useful to the very big players, the top ten, let’s say, in the in the market, 3PLs, brokers, forwarders that have reached a certain plateau, let’s say, and cannot grow double digit anymore because obviously they they are already very big, very complex, and this is opening again new doors. Like if you didn’t have before the management didn’t have time yet to look into the unprofitable contracts or to analyze maybe the routes that need optimization or simply having the courage to go to new markets, you can learn a lot about it, about the restrictions, about the road taxes, about everything. This is becoming now very possible, very easy with this kind of technology. That’s why I’m saying that it’s very clear for me how the logistics landscape looks now. It’s I’m very curious how it will look like in the in the next two to three years.

Cristina

Thank you very much.