Key Takeaways
What does a freight operation look like in 2030? Felipe Capella, CEO and co-founder of Loadsmart, answered that question in the opening session of Sightline 2027, held at our Chicago headquarters on September 16th, 2026.
His view is that software, AI and freight experts each own a different part of the work. The system of record holds the truth about every load, AI agents handle the repetitive exceptions, and freight experts make the judgment calls an AI agent can't.
This post breaks down his session, "Software, AI, and Humans: The Freight Operation of 2030," including where Loadsmart AI fits today and how enterprise shippers can start with one workflow. The full recording is below and free to watch, and Felipe's deck is further down the page.
Sightline 2027 is Loadsmart's annual event for enterprise shippers, and the first edition took place on September 16th, 2026, at our Chicago headquarters. It carries the 2027 name because each edition looks one year ahead, so this one was our sightline into next year. For a look at the full day, including the customer panel with Babylist and Scotts Miracle-Gro, read our Sightline 2027 recap.
Most enterprise shippers today have software that records their freight and flags problems, and a person still has to go fix each one. It's estimated that most enterprise freight will move past that by 2030, to an operation where AI agents complete the work inside rules the shipper sets and freight experts handle what's left. Felipe compared it to self-driving cars, where the car goes from warning the driver to doing the driving itself.
Loadsmart AI runs that way today. A shipper can have AI agents completing work inside its guardrails within three months of starting. Felipe's closing question to the room was why wait until 2030.
Each of the three does a job the other two can't, so an operation that drops one of them runs into a blocker. Felipe described it as a tripod.
Software is the system of record. That's the ERP, TMS, WMS and YMS, the single source of truth for what happened to a shipment, who touched it and which documents are attached. Felipe expects these systems to stay. AI agents work on top of them through the same channels people use (email, phone calls, chat, and computer use in a browser), so a shipper keeps its current stack and skips the long integration project. Loadsmart's own systems of record, Opendock for gate, yard and dock and ShipperGuide for TMS, sit in that same layer.
AI is the execution layer. Felipe described the daily work of freight as thousands of small compliance gaps. His example was a commodity that requires a carrier status update every six hours. When eight hours pass without one, someone has to call the carrier, get the update and log it, and that happens across a large share of loads every day. Some large shippers have teams of 200 to 300 people doing this work. He grouped most of it into six categories, which match the six Loadsmart AI agents available today:
For workflows outside these six, our team can scope building a custom AI agent for your team.
Freight experts own the judgment and the relationships. Depending on the workflow, AI agents handle 70 to 90% of these tasks, and the rest route to people. That can be the shipper's own team, or Loadsmart's operations team if the shipper wants the full outcome handled. Felipe gave two reasons people stay in the loop. AI still has limited context on the physical world, so some exceptions will always need human judgment, and freight runs on trust with carriers, vendors and customers, which people build and manage.
What Do Enterprise Shippers Get Back?
The main benefit Felipe named is time, for shipper leaders and for everyone on their teams.
When software, AI and freight experts run as one system, the payoff is that repetitive work gets done faster and at a lower cost than doing it by hand, which frees up hours across the team every week.
Felipe gave an example of where that time should go. Instead of making 30 phone calls to chase PODs, someone on the team could look into why freight from Tennessee to Illinois went up 20% last quarter, which could be worth $2 million in savings. He also tied it to attrition, since the calls and compliance checks are the work people tend to dislike most, and removing them gives teams back their nights, weekends and holidays.
Pick one workflow and test Loadsmart AI on a limited set of loads. That can be one of the six categories above or a custom workflow, for example a scheduling process a team runs by hand today.
Loadsmart is offering three ways we can showcase what Loadsmart AI can do with your freight, so you can see it in action first:
The goal is a working proof of concept in about two to three months, so a shipper who starts this fall could have an AI agent running by the end of the year. AI runs your freight, and our experts handle the exceptions.
The case for starting now is that AI adoption in freight will only speed up from here. Felipe compared AI adoption to the internet in the 90s, a trend that kept compounding once it took hold, and said the enterprise shippers who start with one workflow today will be the ones who have built their AI agents by the time everyone else starts testing.
Sightline 2027 was an exciting step in exploring the future of freight and AI with the enterprise shippers building it alongside us. The future of freight is being built now, and we're already planning Sightline 2028 to keep the conversation going.
Every Sightline 2027 session is on the Sightline 2027 hub. That's also where you can request the personalized demo, onsite workshop, and proof of concept.
Felipe Capella expects most enterprise freight to run with AI agents completing the repetitive work inside rules the shipper sets by 2030. Systems of record like the TMS, WMS and ERP stay in place as the source of truth, and freight experts handle the exceptions and relationships that need a person.
AI agents work on top of a shipper's existing systems of record, so the TMS, WMS, YMS and ERP stay in place. They use the same channels people use, including email, phone calls, chat and computer use in a browser, which means a shipper can start without a long integration project.
Enterprise Shippers can start by picking one workflow and testing Loadsmart AI on a limited set of loads. Loadsmart offers a personalized demo, a one-day onsite workshop and a proof of concept on the shipper's own data, all before any contract starts, and the goal is a working proof of concept in about two to three months.