The 2030 Enterprise: Freight Execution Per Human Decision

Early in my career, my whole job was typing receiver notes. When the dock opened, which door to use, who to ask for, how long you’d actually sit. I was the company’s memory, one warehouse at a time, at the speed a person types.

Then freight spent 20 years buying software to help that kid type faster. I bought plenty of it myself, and I’d like most of that money back.

Now I run Envoy, and we built Ellie to do the typing. She works every phone and inbox at once, negotiates inside your guardrails, and checks MC and DOT before your rep even looks up. About 75% of freight in our live accounts books through her.

That leaves your people with the only work worth paying them for: the decisions. How much work each decision sets in motion is what I call freight execution per human decision, and it’s the number the 2030 enterprise will run on. 

Most brokerages have never measured it.

The Software Paradox explained why software never shrank the work, and Beyond SaaS moved digital labor onto the payroll. This one hands you the scoreboard, and I promise it won’t flatter you.

What Does 4:30 on a Friday Say About Your Brokerage?

Just about everything. A carrier falls off a reefer load out of McAllen, the customer has already texted twice, and your best rep drops what she’s doing. She sends 40 emails and makes a dozen calls to scrape together three offers. She books a truck $400 over what she wanted, updates the TMS, and texts the customer. Then she makes her kid’s soccer game in time for the second half.

Now count what she genuinely decided: which carriers to chase, how high to go, and which truck to trust. She made maybe three real decisions, buried under an hour of execution a machine could have handled while she was still reading the first text.

That’s what freight execution per human decision measures: how much work gets done every time a person has to step in. On that Friday, she had to step in at every turn, so the number sits around one. Your best rep spent the afternoon as her own assistant.

Why Did Freight Ever Start Bragging About Headcount?

Because for 20 years, the only way to add freight execution was to add people, and I’m as guilty as anyone.

Every brokerage I ran paid the same price for growth. Win freight, hire reps, add a manager, then spend months getting the new floor to perform like the old one. At 25, I thought a fuller floor meant a better company when, really, it mostly meant a bigger payroll moving the same work by hand.

The industry still does it. Somebody announces they’re “now 400 strong,” and everybody claps. Four hundred strong at what? If half those seats exist to retype pickup times into the TMS, that press release is a confession.

And please stop using the tiresome “freight is a people business” cliché as a reason to keep it this way. I agree it’s a people business, which is exactly why it kills me to watch great people spend their day as human copy-paste.

What Happens When the Machine Does the Typing?

I’ll break it down this way. We went live with a new customer recently. On the first load they posted, Ellie took 76 inbound carrier calls in five minutes, all at the same time. She pulled 20 offers, vetted every carrier through Highway, and booked the load below the posted rate.

On a normal floor, somebody answers eight of those calls, books one, and the other 68 never existed.

Now rerun that Friday. The carrier falls off at 4:30. Before your rep finishes reading the customer’s text, Ellie is already working the phones, pulling offers, and checking MC and DOT. Your rep makes the three decisions that need her and still catches kickoff.

Can’t You Just Buy an AI Tool and Call It a Day?

You can, and most executives think that’s the plan. McKinsey surveyed more than 10,000 of them for State of Organizations 2026, and most expect AI to stay mainly a support tool in the near term. Buy that version, and you get approval theater.

The tool drafts 300 emails, your rep clicks “approve” 300 times, and the boardroom calls it an AI strategy. Your number didn’t move. You invented a slower way to send spam, and your reps proofread chatbots dressed up for the sales meeting.

Worse, those clicks stop meaning anything. Gartner warned in May that human approvals degrade under time pressure and approval fatigue. But I didn’t need Gartner to tell me that. I’ve watched reps approve things at 4:58 on a Friday without reading them.

So we built Ellie the other way around. She shows her work, pauses for approval, and earns more rope as your team trusts her. Your people keep the high-stakes calls, starting with carrier selection, because after Montgomery v. Caribe you may have to defend every pick to a jury.

What Do Your Best People Do All Day Now?

They finally get to use their heads.

On one live floor, Ellie told an operator a delivery appointment couldn’t move, and the TMS agreed. He overrode her anyway. He knew that the receiver takes freight Friday or Saturday, first come, first served, and nobody had ever written it down.

That’s a receiver note, the same kind I used to type. In the old model, it solves one load and goes back into his head until the day he quits. With Ellie, it lands in the Carrier Context Graph your brokerage owns, and the next load to that dock doesn’t need him. You only train a digital worker once.

I’ve said the traditional carrier rep role won’t exist in the U.S. by 2036, and I bet my company on it. A brokerage with 100 reps today still needs about 100 people in 2030, deciding instead of typing. When my own job flipped, it felt like riding a bike downhill after a lifetime of pedaling uphill in the mud.

Who’s Already Running the 2030 Playbook?

The biggest names in brokerage, and they’re telling Wall Street about it.

RXO told investors on September 16 that its brokerage headcount is down a mid-teens percentage from a year ago while volume grows. It says it could handle about 15% more freight on today’s staff, and it still keeps a human in the loop. That’s the 2030 enterprise showing up four years early.

Smaller shops are moving too. Michael Cherney, CEO of Cooler Logistics, told FreightWaves in July that his team has scaled without adding headcount since deploying Ellie, even as the market tightened.

And the market isn’t handing out grace periods. FreightWaves’ spot index hit $3.44 a mile on September 9, up 47% in a year. In a squeeze like this one, whoever gets their number up first sets the cost per load everyone else has to match.

What Should You Do Before F3?

Count one load this week, then come talk to me.

This is my third post in a row arguing that the software-first model is done. The Boss of Machines named the new job, Digital Operations named the discipline, and the last two posts made the case for a digital workforce. Freight execution per human decision is how you keep score.

Book 30 minutes and bring that load with your count. We’ll run it live in your own portal, and you’ll see your number with Ellie on the desk before the call ends.

Rather hear it from someone who isn’t selling anything? Ask for a reference call with a brokerage already running her. Or email me at robby@tryenvoy.ai with the load and your count. I’ll answer myself, probably faster than your TMS loads.

I’ll be at F3 in Chattanooga on October 27 and 28, arguing that the software-first operating model has hit its limit. Last year I walked that floor and saw a lot of old software in new AI jackets. Find me this year and bring your number.