Alvys opens freight AI agents to fleets of all sizes

To understand the value of agents, imagine a trucking company that runs two operations at once. The one most known moves the freight. The other proves that freight moved on time, safely and under the right authority. The second involves all the paperwork, and in a post-Montgomery world, may decide who wins the next contract.

Large fleets have the staff for the second operation. They employ legions of specialists to watch federal safety records and dispute shipper service failures before either hardens into a number a customer can see. Small fleets find out what they missed after the damage is done. That is the gap Foundry is aimed at, and Darman told FreightWaves that experience came from personal experience.

The DOT knock that started it

Darman started a trucking company in 2004 while at Georgia State University, spent a few years at J.P. Morgan, and went back into trucking in 2010. Growing that business, he hit a form of fraud he did not know existed.

“Drivers and owner-operators would fraudulently use my signs, my DOT number, my authority, and run freight without my consent. They were racking up violations under my name. I had no idea until a year later, when the DOT knocked on my door for an audit.”

He had 10 to 20 drivers then and knew all of them. Logging into the Federal Motor Carrier Safety Administration portal every morning never occurred to him. Larger carriers have the resources to submit a Request for Data Review (RDR) to the FMCSA. The request is accessed through the FMCSA DataQs system and requires carriers to prove they were not at fault for the crash. It is an ongoing process to submit, process, verify and track DataQs requests — something smaller fleets and owner-operators are most at risk of but least aware of.

“You’re actually being labeled as an unsafe company. But that’s not necessarily the case,” Darman said. “When you look at how some of these owners or entrepreneurs run their companies, they run their companies very safely. But the underlying data on FMCSA doesn’t support the way they run their companies.”

Freight AI agents versus the 24/7 watch desk

The industry’s fix for that problem is headcount, available only to companies that can pay for it.

“They have employees on it around the clock. They log into FMCSA every day to make sure that there’s nothing shady happening,” Darman said. “And when something is happening where a driver that’s not employed by the carrier is being reported on FMCSA, they dispute it right away.”

Foundry replaces the shift with an agent.

“You could  have an agent check an integration with FMCSA,” Darman said. “The moment something doesn’t match the driver records on file, it flags the owner. That way the owner can dispute the record immediately.”

Darman was direct that the use case is still theoretical: “I haven’t used that with anyone yet, but I think it’s such a powerful use case.” The argument behind it is a cost line. “Instead of having employees doing this work for you, you could actually deploy this agent that’s not costing you much,” he said.

On time and on the scorecard are two different things

The same asymmetry runs through shipper relationships.

“Being on time for every pickup and delivery is one thing. Making sure the shipper’s scorecard shows you were on time is another. ,” Darman said. “Those are two different things.”

Small carriers argue the facts, but without data, shippers fall back to their records. “They’re just arguing, ‘Hey, I was on time,’ but [a large shipper] will look at you: ‘Well, I’m sorry. My scorecard shows otherwise,'” Darman said.

Large carriers have treated scorecard maintenance as its own discipline for years, flagging EDI 214 status messages that land outside a delivery window and routing them to customer service teams that recategorize the miss under weather or another excusable code. It is the staff, and not always the service, that shows up in the numbers. Inflated carrier scorecards are an underreported aspect for shippers managing their carrier partners’ performance.

For those disputes, speed often decides whether the dispute lands. “If you do it in real time, the shipper remembers and will always take that off the scorecard,” Darman said. “If you do it too late, then you have a problem.”

Context is the product

Foundry’s claim to differentiation relies not on the agents. Instead it focuses on the operating data underneath them.

“We have the freight context and we understand your lanes. You don’t need to keep up different integrations,” Darman said. “We have the context and therefore there’s no extra login.”

The platform sits on Alvys’ existing data infrastructure, more than 120 integrations, native EDI connections to hundreds of shippers, and a SOC 2-compliant security foundation. Enterprise agreements with model providers keep customer data out of public model training.

Governance ships with it. Agent Shield logs agent activity, keeps actions traceable and routes high-impact decisions for human approval. An intelligent model picker selects the cheapest capable large language model for each task, a hedge against compute costs climbing with usage.

Who holds the keys

Most AI providers entering logistics write the agent logic themselves and keep the operating knowledge in-house. Alvys is handing operators the building blocks and training their teams to assemble agents themselves.

“I lived the other version of this as a trucking company owner,” Darman said. “Vendors got me started, then charged for every customization, every report, every integration. They held the keys, and leaving was a struggle. We are building Alvys Foundry so our customers hold the keys instead.”

One early customer reads the distinction the same way. “Foundry is not AI for show. It is AI with a job description,” said Carlos M. Llanes Jr., founder and CEO of Spartan Carrier Group. “It helps remove the manual friction from freight so our people can stay focused on judgment, service and elite execution.”

Foundry was first shown at a June 17 Customer Advisory Board meeting, where the first customer cohort filled. The waitlist for the second is open.

What separates operators from here, in Darman’s view, is no longer the technology.

“Tech is not going to be the limitation,” he said. “It’s the deep thinkers behind the organization who will set up the SOPs in such a way where you’re going to be the winner. And I believe that AI will now expose who the real workers are.”