Fleet Owners Are AI Suppliers Too
Quick setup before the fun part.
what an FDE actually is
Everyone has decided to call their consultants "forward deployed engineers" now. Fine. Words are free.
Here is the real definition, and it is one sentence:
An FDE's deliverable is a working outcome inside the customer's environment. An FDE's byproduct is a product.
That second clause is the whole thing. If the byproduct is not a product, you do not have an FDE. You have a contractor.
Test it yourself. Does deployment two cost less than deployment one? If no, nothing is compounding, nothing is being absorbed into a platform, and you are running a staffing agency that tells itself a product story at the off site.
The part about successive deployments costing less means nothing to you if you are the first deployment, unless you benefit from the rest. Park it. We will come to it.
The industry mumbles through data ownership
The client's data is what makes the platform valuable. Twenty years of lane performance, maintenance failures, detention fights, driver churn. All of it paid for in real money and real mistakes.
That data goes in. It trains models. Those models get deployed at thirty other companies, some of whom compete with the client for the same freight.
The client's compensation for this? A better workflow and an invoice. Better EBITDA? Not if all the workflows aren’t overhauled.
Meanwhile the vendor is building a $200M asset out of thirty clients' hard-won institutional knowledge, and the client's claim on that asset is exactly zero.
Nobody says this out loud in the sales meeting. Everybody thinks it in the parking lot afterward.
The standard vendor answer is "our DPA has strong anonymization provisions," which is a legal answer to an economic question.
We solve it with participation
We stopped arguing about ownership. Ownership is a lawyer's question. We moved the fight to participation, which is a business question, and our buyers are business owners.
Three pieces:
- Vertical IP that exists before we knock. We arrive with fleet-specific models already built. The client work is not the R&D budget. They are deployment N, and deployment N is cheaper than deployment 1. That is the only honest proof that we are running an FDE model and not a body shop. If they are deployment 1, the incentives are worked out differently.
- Upside priced against valuation unlock, not hours. We participate when the enterprise is worth more. Not when the project is "delivered." Delivered is a vendor word.
- Distribution share plus a seat on the cap table. If a client brings us into their network, they take a share of what comes through that door. And they stay on as an advisor with equity, which means the data tension does not get resolved by a contract clause. It gets resolved by putting them on the same side of the table as the asset their data is building.
That last one sounds soft. It is not soft. It is the single most commercially useful thing we have done, and I have proof, which is the actual point of this post.
Real case. Made up numbers.
Dale ran 340 trucks for 22 years. Dry van, some reefer, Midwest regional. His dad ran it before him with 40 trucks and a fax machine.
He is our advisor. Now he has equity and he has a phone full of people who ran fleets alongside him for two decades.
Last week he got on a call with Mitch, who runs about 400 trucks out of Ohio. Dale set it up. Dale ran it. We were on mute.
What follows is paraphrased and the names are changed, but the motion is real.
the call
Pleasantries were exchanged. Small talk about the industry happened and we understood nothing. We were introduced. Lets get to the meat of it.
Dale: These guys took my data. All of it. Twenty two years of lane performance, every detention fight I ever had with Menards, every maintenance record including the four hundred that just say "fixed it."
Mitch: And?
Dale: And it made their product better. Permanently. For everybody, including guys who bid against me.
Mitch: So why are you on this call?
Dale: Because they were the only ones who let me argue about that instead of handing me a data processing addendum and changing the subject.
Mitch: Everybody says they're different.
Dale: Sure. So ask them three questions and you'll know in ten minutes.
One. Ask what they've already built for trucking. Not "we're a horizontal AI platform and logistics is a great vertical." That sentence means you are the R&D budget. You want a noun with a number attached to it. When they came to me they had a detention model built on stop level data from fleets that weren't mine. That's the difference between deployment one and deployment nine.
Two. Ask what the next fleet pays. If it's the same as what you pay, nothing is compounding, and in year three you're still paying full freight for work they already did once.
Three. Ask what breaks first. If they don't immediately start talking about your data being garbage, they've never deployed anything anywhere. Mine was garbage. Yours is worse, no offense.
Mitch: Some taken.
Dale: The guy who came to see me said "your maintenance data is going to be worse than you think and you won't see anything for six weeks." I almost threw him out. He was right.
Mitch: Okay. What's the catch on the money.
Dale: There isn't a rate card in the way you're expecting. They price against what your business is worth after, not against hours in. Which sounded like consultant poetry to me until my CFO worked through it and realized they only get paid well if my multiple moves, not if a project gets "delivered."
Mitch: And you're on their cap table.
Dale: I am, and you should assume I'm biased, and I'd rather you assume that than find out later.
Mitch: Appreciated.
Dale: But run the arithmetic on why that's there. My data built part of their asset. So instead of paying me in dashboard credits, they put me in the asset. That's not a favor. That's them recognizing that guys like us are a supplier of a critical input, and suppliers get paid.
Mitch: And if I bring them into my group?
Dale: Then you take a share of that too. Which, between you and me, is the most underpriced thing either of us owns. We both know every fleet owner in three states and nobody has ever paid us a nickel for it.
Mitch: What do I actually do first.
Dale: Ninety days. One P&L line you already fight with your CFO about. Deadhead or detention, pick one. Baseline agreed in writing before day one, because everybody discovers a baseline after the fact.
And then the only test that matters. Week ten, is your dispatcher using it without being told to? Mine is named Denise and she killed four vendors before this one.
what we learned watching
The advisor can say things we cannot. "These guys took my data and it made their product better" is a disqualifying sentence from a vendor and a credibility bomb from a former operator. Same sentence. Completely different reaction.
He led with the objection. We lead with the solution, because we are proud of it, which is a founder disease. He spent the first three minutes on the thing he doesn't like. That is what bought credibility.
He translated our model into arithmetic, not adjectives. We say "aligned incentives." He said "they only get paid well if my multiple moves." Same idea. But coming from another operator, it came across as endorsement of value.
The cap table disclosure was an asset, not a liability. He flagged his own bias in the first third of the call. Every prospect assumes bias anyway. Naming it converted a hidden discount into visible credibility. Besides, every fleet owner can be a part AI owner. That removed all conflicts.
Denise is the market. We have been benchmarking model performance. He benchmarked dispatcher behavior in week ten. He is right and we have changed how we scope pilots.
the actual point
The vertical IP is why we can deploy fast. The FDE model is why it lands in the workflow instead of in a dashboard. Those two things are replicable and someone will replicate them.
The advisor on the cap table is the part that compounds, because it does not scale linearly with headcount and it does not show up in a competitor's product roadmap. It shows up as Dale calling Mitch.
We spent years building the first two.
The third one is what closes the deals when we watch while on mute.