Anthony Fioritto II was sitting with James Parker Jr. and other senior executives from Stellantis, making the case for planning and transport systems he says have been on the market “for a minimum of 20 years, various degrees of sophistication”, when he looked out of the window and watched an Amazon drone go past.
“There goes a robot that has now set the expectations that you can order a pen, and it will drop it off at your house in one hour,” he says. “That is exactly what we’re up against.”
Fioritto, vice-president supply chain North America, for parts and services, works alongside Parker, who is responsible for Stellantis’ aftersales supply chain and logistics globally and across North America. For Fioritto, the issue is less about the technology itself than the pace customers now expect. The five-year aftersales transformation Parker set out to Automotive Logistics last year is going to plan. What has changed is the standard it is measured against.
A plan that cannot be set in stone
“Everything is on track, on schedule as we designed,” Fioritto says, allowing for “minor, day-to-day hiccups”. The mega hub in the Detroit area remains scheduled for next year with its timing secured, the Forsyth parts distribution centre (PDC) in Georgia opens in the coming months ahead of schedule, and the forward stocking locations extending the network “have been incredibly well received and have brought in additional revenue based on parts availability”.
The harder part is that a five-year plan has to survive several years of technology shifting underneath it. Fioritto describes second-guessing himself in year two and year three, as new technologies emerge and some prove more useful than others. “A plan like this can never be set in stone and then that is what it is and there’s no change,” he says. He has sent a robotics company into one of Stellantis’ legacy PDCs to reassess what the operation is trying to achieve, expecting that the findings might reset part of the plan.
Any change would have to be justified against money the business has already committed. “We’re talking about big money. It’s difficult to go back to the company and say, you gave me this money and now I want to use it for something else.”
Getting everybody close to the sun
Fioritto’s biggest challenge in the transformation is neither a system nor a site. It is communicating what the transformation actually is, down to shop floor personnel. “I don’t think everyone truly is able to understand the magnitude of the change that is happening and what their role is in better serving the customer.”
His answer is what he calls “the industrialisation of the customer”. Some groups sit “really close to the sun”, he says, referring to how directly they can see or measure customer service, while most of the organisation never feels the heat – that is, they are several steps removed from dealers and service centres. The objective, he says, is to see “in every leg of the supply chain, all degrees from top to bottom, people realising what the customers are suffering and then also what they are enjoying.”
In practice that means walking into a warehouse and seeing a visual management board or a monitor showing how many customers are distressed, whether they are waiting for parts or have received parts that failed them. He wants the same at the supply base and with the carriers, where that awareness rarely reaches the driver. “Not all of the drivers go with the bill of lading or the manifest and say, I have 10 frustrated customers I’m driving parts for today.”
Few supply chains manage it consistently, he says. “There is no supply chain in the world that has what I’m describing done expertly. They only do it in pockets.”
Fioritto has started to put the idea into practice. He used AI to generate images of what such boards would look like on a shop floor, then took them to colleagues at parts warehouses to argue that without the visual there is no urgency. His argument is that the 10 steps an operator takes today are carried out at a regular pace, but that if a champion is assigned and the waiting customer is made visible, those steps get quicker. He is now recognising the best-performing Stellantis part centres for customer response, has set reporting metrics around it, and has taken the same exercise to a supplier event, setting out six or seven things he wants to see before working through a base of more than 4,000 suppliers one by one.
All the software in the world, he argues, “won’t matter as long as human beings are involved in doing the actual transaction”. If the system understands the customer issue but the people carrying out the order do not, the problem has moved rather than been solved. “It just resides in the AI in terms of the customer issues and it’s not actually happening where the transactions occur by humans.”
Fioritto extends that challenge to technology providers. He wants a measurable indicator of how many people received customer information and what they did with it. Without one, “your system is only as good as itself inside of its own little ecosystem”.
It is the same demand that other automotive manufacturers have been making of their own systems and organisations, including Autoliv Americas, shifting IT from being a service function to a strategic partner across the supply chain.
What Stellantis wants AI to do
“Our biggest challenge right now is the integration of AI,” says Parker, appointed global senior vice-president of supply chain, for parts and service, in autumn 2025. The first application is in updating its transport management system (TMS), where Stellantis is working to give dealers better ETAs and tracking on inbound material. Recalls and wider industry quality issues have put pressure on dealers, who are scheduling recall work alongside warranty repairs. “For them, it’s really critical to understand when the parts are going to arrive. And we haven’t been able to do that really well yet.”
The second is demand and planning. Stellantis carries significant safety stock because the alternative can be more costly, and some of it eventually becomes obsolete. The forecasting approach is an area that still needs work, with the objective to drive more detailed demand and supply planning down to part levels. “The way we forecast is not part by part, detail by detail. It’s in categories. And we really need to be able to do this stuff part by part,” Parker says.
He is also looking closely at what automation could mean for the network. He recently toured one of Amazon’s two most automated facilities, near Boston, where robots play a major role in the operation, and considered what some of those approaches might mean for Stellantis, particularly in terms of cost and maintenance. “There’s certainly things that we can learn and take from that.”
Fioritto is considering the same question in an environment where changes to people’s roles and responsibilities need to be managed carefully. Nearer term, he is pursuing the advanced supply planning and TMS platforms peers have already adopted, and developing a virtual, AI version of a PDC manager. With two, three or at most five people in PDC management, the intent is to take daily negotiation out of the job by reading order priorities, who is at work and what equipment is available, then issuing a plan for the shift. He is starting small, and cautions against assuming the technology will do the work on its own. “You’d be mistaken to think that technology like AI will just make your life completely better.”
One network, several fulfilment models
Parker returns to the Amazon comparison from the other end, as a customer who ordered something recently and had it four hours later. “Customers are getting more demanding. They want more real-time material. They want it not only every day, they want it within hours. So we have to adjust to that, and that requires us to change the way we do business altogether.”
E-commerce brings requirements the traditional dealer flow never had, including UPC labels on parts and master packs broken down so that a dealer who used to receive a case or a pallet now receives, in Parker’s phrase, “onesies, twosies”, alongside truckload volumes to wholesalers. That variable order profile creates inventory discrepancies and compresses planning time, made harder by long lead times on material out of China, where most OE suppliers sit. “You really have to be on point with your planning, which is why we want to integrate some of these AI tools.”
He sees room for the logistics market to evolve further, particularly as newer entrants bring different approaches to delivery. Stellantis has been testing whether on-demand models could serve smaller dealers. Parker uses DoorDash as an example, with the appeal being more immediate access for dealers ordering at any point in the day. Running full truckloads to wholesalers and single items to smaller shops means funding two delivery methods at once, and a couple of pilots are under way before anything scales. Labour remains another consideration, with Fioritto reporting that a supplier had returned to request additional funding to help cover the ongoing driver shortage.
The next test is reaction time
Reacting faster does not mean running one model everywhere. “We’re customising according to the region,” Parker says. Europe brings its own considerations, with policy and dealer communication varying between countries and cross-border movement a constant part of the operation. Parts cannot pass directly from France into Algeria, so that flow goes through a third party.
Underneath the geography sits a timing challenge. It takes about 36 months to plan a product, from building the vehicle through to preparing the parts, work streams and flows. “Now we’re being asked to do this in half the time. And sometimes 12 months or less.”
That shortening cycle can create pressure in aftersales, too, with backlogs, backorders and parts sometimes missing at launch. New models arrive with no sales history, so volumes must be estimated. “You’re guessing in a lot of ways,” Parker says, which makes it important for the organisation to spot when reality starts to differ from the plan and adjust quickly. “If you have antiquated systems and you’re not looking at this real-time on a day-to-day basis, you have delays. And now it takes you months to recover when it should take you weeks.”
The gap between weeks and months is where the two arguments meet. Better ETAs, part-level forecasting and a virtual PDC manager shorten the time it takes to see a problem. Whether recovery then takes weeks depends on whether the people picking, loading and driving understand what is waiting at the other end. Fioritto, who notes that some other supply chains are further along, knows where the challenge lies. “It’s the effort and sustainment behind it that matter.”
