1. How is AI changing how you manage a fleet in trucking?
AI allows fleet owners and operators to process and generate massive amounts of data in ways that were not possible before. This means that it has become feasible to automate new processes and generate real-time insightsv that allow for faster, better decision making.
One example of this could be optically perceiving and reading documents that are common in the trucking world, but rarely have unified formats (BOLs, scale readings, etc.). An AI can ingest these unstructured documents, extract useful information, and feed it into the systems that run the fleet. By using this technique, managers and drivers save significant amounts of time when processing emails and workflows.
Another example is how AI can help real-time decision making in fleets. This can be by surfacing useful information upon prompt or providing predictive analytics, such as better ETA estimates (learn more about an AI-powered (TMS with predictive ETA). It can also help dispatchers in making driver or equipment assignments that improve service level or reduce costs or whatever other goal they would like to optimize for.
2. What does an AI-first fleet look like?
An AI-first fleet has an underlying data infrastructure that allows drivers, vehicles, and managers to generate and capture information for recordkeeping and decision making.
In this kind of fleet, all vehicles are equipped with ELDs, the drivers use tablets or phones for their on-field workflows, and there is a centralized system, usually a TMS, that taps into all data sources and systems (ERP, accounting software, FMS, etc.). Data flows into a central repository where it can be processed and analyzed by AI that orchestrates the operations and automates processes that had to be done manually by teams of human operators in the past.
Note that not only is the richness of data important, but it is real-timeless (low latency) as well as the availability, scalability, and cost efficiency of available computational resources that AI algorithms often need.
Today, 99% of all transportation management systems, especially large enterprise legacy systems, struggle with delivering the data capabilities an AI-first fleet requires. Only certain modern TMS like BeyondTrucks are built to facilitate the real time data and processing requirements an AI-first fleet requires.
3. How can AI help reduce empty miles in logistics?
Empty miles are one of the big inefficiencies in truck transportation and logistics. They represent underutilized capacity as well as poor utilization of assets and driver resources, but in differently they represent wasteful cost.
AI can be leveraged in multiple ways to reduce empty miles. Firstly, at a strategic level, AI powered network analysis can help logistics operators identify markets where filling empty miles is best done with contract business versus spot freight.
Tactically, AI is helpful in improving predictions of ETAs and PTAs so that operators can decide reliably whether there’s enough time for a carrier to complete a backhaul or whether taking on a
backhaul may create a risk of the carrier not being back in time and ready for headhaul contract business. Many carriers today still rely on the stability of the contract business on the head haul and without better.
ETA/PTA reliability may forfeit the opportunity to load the truck for the backhaul and return to the home location empty.
4. Can AI automate dispatching for private fleets?
Dispatching private fleets can be automated both with AI and classical techniques such as optimization algorithms. The paradigm shift we have seen recently is in the availability of data – i.e., how accessible the data to automate dispatching has become within an organization.
If all data about driver and equipment availability, pickup and drop off locations, loads, etc. is gathered in a single database, a machine can make decisions about how dispatching should be organized.
It is possible to use machine learning algorithms, trained on past dispatcher decisions, to automate all the decision making. The limitation with that approach is that dispatch decisions will only become as good as the dispatchers that generated the training data.
However, we have known an alternative way to solve this problem for decades, using classical optimization algorithms. Using this technique, a computer calculates solves mathematically for the best outcomes for a fleet. The problem with classical optimization is that writing down the math to solve the problem is heavy, which means that these systems are complicated to use. This is where modern technologies leveraging AI come into play: LLMs can allow the user to interact with classical optimization systems in a simpler way, by writing the math for us.
Matias Oberpau is Director of Business Development at BeyondTrucks, where he leads growth in AI-native fleet and logistics solutions. He holds a degree in Industrial Engineering from Universidad Católica de Chile and an MBA from Stanford University. Matias excels in AI-first operations, cloud-native TMS integration, and fleet automation—helping clients cut costs, boost utilization, and build resilient operations.