When AI Makes Your People More Productive, Who Decides How Many You Need?

On May 7, a Morgan Stanley analyst named Ravi Shanker asked RXO's chief executive a question most companies would rather not get. RXO, one of the largest truck freight brokers in the United States, had just told investors that its AI tools had "automated more than 500,000 phone calls in the quarter." Shanker wanted the other half of the fraction. What share of all calls was that?

"It's a very low percentage of the number of calls, Ravi," said Drew Wilkerson. "We're in the very early innings of this," he added.

That answer is why RXO is worth a close look. A count of automated calls says little until someone supplies the other half of the fraction. RXO, based in Charlotte, North Carolina, with $1.4 billion in first-quarter revenue, gave investors enough of both halves to show how a business that runs on phone calls and email actually changes. It put software in its reps' inboxes, counted what each person handles, and said out loud that hiring would grow more slowly than volume. Five months later, a bigger rival agreed to buy it. If the deal closes, the question of how many people RXO needs will pass to someone else.

A business that lives in the inbox

RXO's brokerage reps come in two kinds: those who deal with customers and those who deal with carriers. Some requests to price one-off loads, which the trade calls spot quotes, arrive by email in the reps' inboxes; others are quoted digitally. Phone traffic is heavy too. If half a million automated calls in a quarter were, as Wilkerson said, a very low percentage, the total runs far higher. This is a company that runs on its inboxes and phone lines.

Early 2026 was a hard time to be in that business. Wilkerson told analysts that trucking capacity was leaving the market because of "regulatory changes and enforcement," while demand for goods "remains soft." Jared Weisfeld, RXO's chief strategy officer, said rates across the industry had risen about 20% since the third quarter of 2025. He pointed to "the squeeze in our contractual book of business." He didn't spell out the mechanics, but the likely reading is that prices agreed with contract customers hadn't kept up with what RXO was paying truckers. RXO's truckload volume fell 12% from a year earlier.

The assistant in the inbox

Late in the first quarter, Wilkerson said, RXO rolled out "an AI spot agent in reps inboxes." Weisfeld said the company had "broadly deployed a new proprietary Spot Quote agent," RXO's own tool, which works on the price requests arriving by email.

It wasn't the only change. The number of truckloads quoted online rose 30% from the previous quarter, and truckers sent in about 15% more offers to haul loads online. Automated systems handled the half-million calls. A new AI fraud-protection agent added protection for shippers moving high-risk freight. Weisfeld described the point of all of it in terms of people. They were becoming more productive, he said, and "spending more time with our customers and carriers."

The early results came with an obvious catch. Reps who adopted the quoting tool first, Wilkerson said, "are seeing an increase in volume and gross profit per load," compared with the rest of the brokerage organization. But early adopters are not a random sample. The reps who reach first for a new tool may already be the sharpest ones. Later in the call he offered a different comparison: the people using it were "winning 15% more volume versus what they were winning." That measures the users against their own past results, which avoids the question of who chose the tool. Wilkerson expected everyone else to be "fully ramped up on the tool over the next few quarters."

The number RXO chose to watch

RXO chose one measure of output per person: loads per person per day over the past 12 months, counting the reps who deal with customers and carriers. The company's presentation shows that number up 15%.

Read it carefully. Brokerage volume fell 8% in the quarter, and when another analyst asked about staffing, Wilkerson said brokerage headcount was "down double digits on a year-over-year basis." When the number of people falls faster than the work, loads per person go up whatever the software does. Some of the 15% is probably arithmetic. The figures don't say how much is the tools.

What made the answer worth hearing was the decision inside it. Wilkerson described a pace: as RXO starts to grow, "the rate that we add heads" will not keep up with it. Weisfeld called it the chance "to decouple volume growth from headcount." Answering a question from another analyst, Stephanie Moore, Wilkerson said: "We value the relationship first. We are a relationship business." RXO's top customers, he said, had been with it for 16 years on average.

What the money said

Then there's the profit line. RXO's adjusted EBITDA, a common measure of operating profit, was $6 million in the first quarter, down from $22 million a year earlier, on revenue that barely moved ($1.425 billion against $1.433 billion). The company blamed about $3 million on severe weather, mostly in its home-delivery business, plus higher fuel prices and the contract squeeze. None of that was the AI's doing, and the AI didn't prevent it. Productivity per person can rise in the same quarter that margins shrink, because the freight market sets the price of the work.

The second quarter showed what happens when the market turns. One-off (spot) loads made up 42% of RXO's truckload volume, and truckload volume grew 2% from a year earlier. In its August 6 release, RXO reported revenue of $1.8 billion against $1.4 billion a year earlier, a big jump from the first quarter, and adjusted EBITDA of $40 million against $38 million. On the call, Wilkerson said RXO had "processed five times more spot quotes via email through the agent." He didn't say five times more than when. On hiring, he said: "We're growing volume faster than what we're adding headcount." He also said "this is still a people business built on relationships, and those take time."

A buyer with a cost target

On October 5, C.H. Robinson, already North America's largest freight broker, agreed to buy RXO for an implied $5.8 billion. Its case to investors doesn't rest on anything RXO measured. It rests on Robinson's own playbook: by applying its "proven Lean AI operating model" to RXO's business, it expects about $300 million a year in net cost savings within two years of closing, a figure that also counts savings from shared offices and services. FleetOwner, a trucking trade publication, reported that Robinson runs more than 100 AI agents of its own in its North American truck business. The deal needs regulatory and shareholder approval, and the companies expect it to close in the first half of 2027.

If the deal closes, an owner with a savings target and more than 100 AI agents of its own will set the pace Wilkerson described on two earnings calls: hiring behind volume, and leaning on relationships that "take time." RXO did what a careful company would do: it put the AI beside its sellers, counted their output, and chose its pace out loud. None of that binds a new owner.

That is the lesson for a company like yours. A slower hiring pace is a choice that lasts only as long as you own the decision, and every productivity gain you prove makes the case for whoever wants to go faster.