The average mid-market broker spends $150 in payroll on every load they move.
Run Crane's domestic brokerage on five humans plus our AI crew. The math is on the next screen.
Freight is in its third year of recession. A representative mid-market brokerage in December 2025 was losing roughly $19 on every load moved, before financing costs. Most of that cost is human. WakeTech.ai removes the human from the routine work and leaves the humans on the work that needs them.
A 40-person domestic brokerage in 2026 does not work.
These numbers come from a December 2025 FreightWaves analysis of a representative mid-market non-asset freight brokerage. They describe one illustrative operation, not an industry-wide statistical average, but every operator running a brokerage of that size recognizes them. Citations and source link at the bottom of this section.
Per load economics
"In brokerage, margin per load is destiny. And right now, for many, it isn't high enough." — FreightWaves, December 20251
Per load economics
Same revenue. Same gross margin. The labor side of the cost stack collapses. That delta is the deal.
C.H. Robinson hit a 31.3% adjusted operating margin in Q3 2025, largely on AI-driven automation handling over 3 million shipping tasks.2
While the rest of the industry is bleeding $19 per load, the top of the market is widening the gap with AI automation. The forwarders and brokerages that move on this in 2026 keep their margin. The ones that do not get acquired or wound down by the operators who did.
The industry has been in a structural downturn since 2023.[3]
One in four trucks is scrambling for loads that don't exist.[3]
Outbound tender volumes fell ~18% YoY by November 2025; long-haul down ~30%.[3]
- [1] FreightWaves, "How are Freight Brokers Staying Afloat?", December 2025. freightwaves.com/news/how-are-freight-brokers-staying-afloat. Per-load figures describe a representative mid-market non-asset brokerage doing ~$30M annual revenue across ~16,000 loads with ~20 staff. Cited as illustrative of mid-market unit economics, not as an industry-wide statistical average.
- [2] C.H. Robinson reported Q3 2025 adjusted operating margin of 31.3%, attributed in analyst coverage to AI-driven automation handling 3M+ shipping tasks. See industry analysis at keynnectlogistics.com and C.H. Robinson Q3 2025 earnings release.
- [3] American Transportation Research Institute and FreightWaves market data, as cited in the same December 2025 FreightWaves analysis [1] and industry coverage of the 2023-2025 freight recession.
A note on the numbers above. The WakeTech.ai column shows the unit economics our own brokerage (MC-340975, operating under SCAC ATGL) runs at today on the platform. Crane Domestic Brokerage's actual numbers will differ based on their lane mix, customer base, and contract structure. The exact targets are agreed in writing during pilot scoping and measured weekly.
Freight just hit its inflection point.
Most of the industry has not noticed yet.
For twenty years freight technology was a race to digitize what brokers and forwarders already did on paper. Load boards. TMS systems. EDI. Tracking. All useful. None of it changed the shape of the work.
The next ten years are different. AI is about to take the routine work out of freight operations entirely.Rate negotiation, carrier sourcing, appointment scheduling, status checks, exception handling, quoting, invoicing, audit. The things a dispatcher does at 11pm on a Tuesday. The things that used to require 50 people for a brokerage of your size.
The forwarders that automate this work first will compound their advantage every quarter. Lower cost per load, faster response times, better carrier pricing, happier dispatchers doing higher-value work instead of data entry. The ones that do not will spend the next decade watching their margins get compressed by the people who did.
This is the moment Crane gets to pick which group it is in.
Shared-codebase SaaS was built for the last decade of freight.
The dominant cloud-native TMS philosophy is one codebase, no customization, configuration only. Every customer runs the same software. Every customer gets the same roadmap. Every customer bends their workflow to match the vendor opinions. This is efficient for the vendor. It is fatal for a forwarder at Crane scale.
Shared-codebase platforms model one operating entity per deployment. No sub-divisions. No regional stations. No hierarchical rollup. Crane is a corporate parent with multi-modal business units and international stations that each run their own P&L.
Shared-codebase platforms are designed around their biggest anchor customer. In the current market, that anchor is a domestic asset-based brokerage. Crane is a global forwarder. Air, ocean, ground, customs. The shapes do not match.
The AI layer on shared-codebase platforms is retrofitted. It sits on top of a data model designed before LLMs existed. You are buying a platform whose AI future is someone else's choice.
Ten AI agents. One operations floor that never goes home.
Each agent is a stateful production service with its own mailbox, its own memory, its own skill catalog, and its own approval threshold. They send and receive real email, work real orders, and earn autonomous authority through demonstrated performance. Click a card to read what they sound like.
We record a dispatcher's clicks and turn it into a new skill in an afternoon.
The Skill Recorder watches a human do something useful, captures the steps and decision points, and saves it as a runnable skill any agent can pick up. Assign it, set the approval threshold, and the agent runs it autonomously the next morning. The crew gets smarter every week without waiting on a vendor roadmap. That is what an AI-native platform actually means.
The system, in detail. And why each number matters to Crane.
Most pitches show you a logo wall. This page shows you the platform. Every number below is current as of this morning, pulled from the live production database and a fresh count of the source code that runs Crane's pilot.
across FreightWake, WakeMail, WakeEDI, WakeDrive and the supporting automation. Production code, written by an operator, not a vendor sales engineer.
From first commit (Nov 2, 2025) to today. An average of 6.7 ships per day. Not roadmap. Shipped.
The real shape of an enterprise freight operation, expressed in a real database. Not 'configurable fields' in a vendor schema.
Across 86 resource groups: orders, carriers, customers, invoices, billing, EDI, tariffs, settlements, AR, AP, fleet, drivers, vetting, and more.
DAT, Highway, Greatwide, FMCSA, NHTSA, NOAA, Google Places, Anthropic, ElevenLabs, Twilio, TomTom, Mapbox, OpenDock, Ryder, TruckerTools and many more. Pre-built.
Every schema change tracked, idempotent, replayable. Not 'wait for the next release.' Apply it tonight, ship in the morning.
$110.7M in carrier pay transacted on the platform. Production volume, not a demo dataset.
Real 204 tenders, 990 responses, 214 status updates, 997 acks flowing through WakeEDI with three live trading partners.
Built by an operator, not a vendor
WakeTech.ai is written and run by an active freight brokerage operator. The founder writes the code, runs the brokerage on the platform, and answers the phone. Every feature exists because an operator needed it on a Tuesday, not because a product manager pitched it in a slide.
A feature a week, indefinitely
270+ commits in the last 30 days alone. Carrier AP, A/R disputes, dunning workflow, settlements reconciliation, WakeDrive driver app, fraud guard orchestrator, skill recorder, DAT bulk posting. Each shipped in the last six weeks. The vendor on the other side of your evaluation cannot match that cadence.
Self-hosted where it matters
Valhalla routing on OpenStreetMap data. Self-hosted JMAP and IMAP mail. Self-hosted EDI processing. Self-hosted weather and corridor intelligence. Zero per-request mapping bills, zero per-mailbox licensing, no metered API surprise at the end of a busy quarter.
Dedicated database, dedicated deployment
No shared multi-tenant database. Every Crane deployment lands on Crane's own SQL Server instance, optionally in Crane's own Azure tenant. Customer data never mixes. A security incident in another customer cannot reach you. Your performance is never coupled to another customer's workload.
This is not a pitch deck. It is running right now.
WakeTech.ai is in production today, handling real freight for a real brokerage under SCAC ATGL. When we pitch Crane, we are pitching a platform we bet our own business on first. Below: what is actually live, with current production volumes.
What ships with a Crane deployment on day one
204 tenders, 990 responses, 214 status, 997 acks. SFTP and AS2. 83,000+ messages processed.
Native driver app. In-app Valhalla truck navigation, NEXRAD radar with forecast, BOL/POD upload, expense capture, Twilio Verify auth.
Highway integration, FMCSA crashes/inspections/SMS, NHTSA recall scan, VIN deactivation policy, email-auth verifier, custom onboarding rules engine.
Age-bucket grid, six inline action forms, tone-graded dunning, dispute → credit memo workflow. By-customer rollup view.
POD-triggered carrier pay sweep, payment routing, invoice gate, settlement lines.
Customer-invoice auto-link from agent settlements. The exact multi-entity reconciliation Crane will ask about.
Bulk post with real-time margin preview. Certification evidence package prepared for DAT.
Skill Recorder captures dispatcher clicks. Time-driven dispatcher, escalation engine, four-layer safety, function-direct runner, recommendations inbox.
Live truck halos on a national map, breadcrumb trails, NOAA radar overlay, Sentinel system state, narrative voice on every meaningful event.
Email-based pickup and delivery scheduling. Four-hour same-day confirm window. Two-step confirm page flow with the warehouse.
Operational consolidation of orders onto a single carrier dispatch. The forwarder shape, not the broker shape.
Per-customer tariffs, lane rates, accessorials, FSC auto-update on a weekly sweep. Intake and Vera quote against the live tariff.
Day 30. Day 90. Day 365.
One Crane business unit runs on WakeTech.ai. The AI crew handles inbound quoting, carrier negotiation, status checks, and appointment scheduling with human approval on every decision. Your dispatchers stop doing data entry and start doing the work they were hired for. The first week feels strange. By week three it feels obvious.
Hierarchical divisions, regional stations, and corporate rollup are built into your deployment. The customization layer holds your customs workflows, your legacy ERP integration, and the pieces of your business that make Crane Crane. Ops reports stop coming out of spreadsheets. Station managers see their own P&L in real time.
Your per-load cost is measurably lower. Your quote response time is measurably faster. Your dispatchers handle more loads per head without working harder. Your corridor intelligence flags disruptions before competitors see them. And the platform is still improving every month because you own it, not a vendor who owns you.
Every line of the P&L.
120 days. One business unit.
Real freight. Measurable outcomes.
No rip and replace. No multi-year migration. No fight with your existing vendor. We pick one Crane business unit where the pain is highest, stand up a dedicated WakeTech.ai deployment, run real freight through it for 120 days, and measure outcomes against the numbers we agree on before we start.
At the end of the pilot, Crane has three things: a production WakeTech.ai deployment scoped to that business unit, a documented case for what it did measured in your own P&L, and the option to expand on terms we agree to up front. If the outcomes are not there, we walk away clean, you keep the data, and Crane has learned something valuable for free.
The asymmetry is the point. Your downside is 120 days and a scoped budget. Your upside is a decade of compounding advantage over competitors who bought the other thing.
I used to work here.
Twenty years ago I was at Crane. They called me Gadget Boy. I left, built a freight brokerage, ran it for two decades, and spent the last year building WakeTech.ai for myself because no vendor was building what an operator actually needs.
I am pitching Crane because I think what works for me will work for a company I still care about, at the scale Crane operates. I am not a consulting firm with a deck. I am the person who wrote the code, runs the brokerage, and answers the phone. If we do this, you are getting the founder on the line for the first eighteen months.
The next conversation is short. Let us have it.