Why Your Last Mile Strategy Fails in the Middle Mile (and How to Fix It)

Troy Lester started in delivery at 18, running food to classmates on meal points. He went on to build micro-fulfillment and last-mile companies before co-founding WARP, a tech-powered freight network that has raised $22 million and now runs 1,500+ LTL lanes, 55 cross-dock facilities, and 7,000+ carriers for retailers like Walmart, Saks, and GoPuff.

In this episode of Logistics Remix, I talked with Troy about the middle mile, the stretch between the fulfillment center and the local delivery station that decides whether the economics work. Here are the takeaways.

Great last mile means nothing if you lose the middle mile

Troy learned this at Jitsu, the last-mile company he helped build before WARP. Jitsu had strong last-mile service and 15 sort centers across the US. The problem was getting freight to those sort centers on time.

Regional carriers couldn’t justify a full truckload to Jitsu’s facilities. LTL carriers couldn’t hit the sort times. What Jitsu needed was per-pallet LTL pricing with truckload-level service, and nobody offered it. Troy’s summary: “It didn’t matter how much cheaper the last mile cost per box was. We were just losing at the middle mile.”

That realization became WARP. The earliest idea was simple: resell cargo vans and box trucks with liftgates, the exact equipment he had struggled to procure for micro-fulfillment sites. Those were WARP’s first 10 customers. Then came the bigger idea: build a network that consolidates freight instead of just moving trucks.

FlowSkip: one truck, one pickup, one dock door

WARP’s signature solution is called FlowSkip. It combines zone skipping and pool distribution into a single pickup.

Here is how it works. WARP sweeps the dock at a brand’s fulfillment center and picks up everything at once: zone-skip parcel volume, store replenishment freight, big and bulky items for local delivery, and LTL pallets. One truck, one pickup, one dock door. The truck runs to a WARP cross-dock in the destination market, drops the B2B and big-and-bulky freight, then hands the parcels to FedEx, UPS, Veho, Jitsu, or whichever last-mile carrier fits. The shipper gets zone-skip economics on parcels and faster, higher-quality B2B delivery from the same move.

This sounds counterintuitive because the industry spent years unbundling. Parcel goes with provider A, LTL with provider B, big and bulky with provider C, store replenishment with provider D. Troy points to the XPO, GXO, and RXO splits, the FedEx Freight spin-off, and the UPS Freight sale as the same pattern. The result, he says, is dock congestion everywhere: four providers, four pickups, rising detention times, and four invoices to audit. FlowSkip re-bundles it.

The network was built by hand

There was no cross-dock network to plug into on day one. Troy and his co-founder visited the first 10 facilities in person, shook owners’ hands, and trained warehouse crews on the app themselves. Year one scan compliance was rough. By year three it hit 100 percent across the network.

They deliberately throttled growth. Instead of going national fast, they built pickup and delivery density in about 10 major metros first, selling to smaller shippers with regional needs. Only then did they connect the long lanes. LA to New Jersey was the first. Today 1,500 LTL lanes run five days a week.

Now the focus is stuffing the existing 55 cross-docks fuller and adding robotics, not adding new facilities. There is a waitlist of warehouses trying to join the network.

“McDonald’s level tech”

WARP built all of its software in-house with the same CTO Troy worked with at Jitsu. Seven years together, same philosophy: technology should make operations so simple and repeatable that anyone can run them. Troy calls it “McDonald’s level tech.” Grab someone off the street and they can run the fry station perfectly on day one.

That philosophy shapes what they did and didn’t build. Troy is blunt: front-end dashboards are overrated. What transportation decision makers need is lower cost and better service, and you can’t get that without technology in the operation itself. So WARP’s TMS understands freight at the piece and SKU level, not just pallet and truck. A routing algorithm prices every proposal. The customer dashboard exists, but the flywheel is the point: lower cost structure, lower prices, more volume, more density, even lower cost.

AI where it actually pays

Troy thinks AI is still overhyped in logistics, but WARP uses it in specific places with measured results.

Customer service: 40 percent of tickets are now answered by an LLM they built and trained themselves. The key detail is they measure satisfaction, one to five stars, against human-handled tickets. Early versions annoyed customers. They kept tuning until the quality held up.

Carrier sales: AI sends lane bids to pools of qualified carriers, and AI now decides which carrier belongs in which pool. Then AI negotiates rates back and forth with carriers. No human, no carrier-sales commission baked into the price. That commission slice just disappears, and the shipper gets a lower rate because of it.

Track and trace: agents flag off-route trucks and late pickups automatically, then send the messages themselves. Because agents misbehave, WARP built supervisor agents that check each step against the overall service commitment.

Forklifts that started with a PlayStation controller

At WARP’s concept sites, automated forklifts unload trucks and stage pallets using computer vision cameras mounted at the top of the facility. The cameras map the building dynamically, which means WARP gets full visibility without integrating with each partner’s WMS. A few sites have eliminated multiple labor shifts this way, with humans still working alongside the machines.

The early testing was gloriously low-tech: Troy and his co-founder drove the forklifts with Sony PlayStation controllers for weeks before the autonomy worked. The vision is fully roboticized cross-docks with minimal human involvement, driven by both safety and cost. Warehouses run temp labor across three shifts, and forklift accidents are more common than the industry admits.

Self-serve is already live for basic quoting: cargo vans, box trucks, 53-foot trailers, and LTL, instant online, no human involved. Next up is self-serve network proposals and cross-dock quoting, plus an AI-powered consultative experience for mid-sized shippers that can generate its own strategic analyses.

The bigger bet

Troy’s framing is deliberately grand. There hasn’t been a great new American supply chain company since FedEx, in his view. UPS is 150-plus years old. His argument: legacy carriers built moats out of terminals and sort centers, then coasted on decades of rate hikes and peak surcharges. Technology is now good enough to match their cost structure without the capital investment.

His equation is simple: more volume equals more density equals lower cost equals lower prices. Quoting Bezos: “your margin is my opportunity.” WARP doesn’t plan to take outsized margins. It plans to keep pushing prices down and let the flywheel spin.

---

This post is based on the Logistics Remix episode “Reinventing the Middle Mile” with Troy Lester, co-founder and CRO of WARP. The views are the guest’s and the host’s own.