An end-to-end automation system for a vehicle logistics operation — from first inquiry to delivered car, without a single manual data transfer.
Lead response time
Driver assignment
Inspection protocol
Saved per month
AutoTransfer moves vehicles across Central Europe — mostly between dealerships, auction houses, and private sellers. A typical week is 60 to 100 transfers, ranging from single cars between nearby cities to batches of ten or more across borders.
Their operation is small but high-touch. Every transfer involves a quote, a scheduled pickup window, a driver assignment, an inspection protocol at both ends of the journey, and final delivery paperwork. Multiply that by a hundred transfers a week and the admin becomes a full-time job — sometimes two.
They came to us because they were growing fast, and every new client made the paperwork heavier instead of the business easier.
We spent two days shadowing the team, watching how requests came in and what happened to them. The pattern was clear:
Email, phone, WhatsApp, Instagram, and a form on the website. Someone had to manually check all five, copy the details into a CRM, and hope nothing slipped.
By the time someone replied, the customer had often already contacted two other logistics companies. Speed-to-lead is the single biggest lever in this industry.
Someone opened a spreadsheet, checked availability, called two or three drivers, waited for confirmations, and then updated the sheet again. Fifteen to twenty minutes per transfer.
Every vehicle needed a condition report, but the formats were inconsistent, and details got missed when transfers were rushed. Disputes later cost more than the transfers earned.
Nobody was doing anything wrong. They were just doing everything manually, and the manual load was scaling with the business.
We built three workflows that share data between them. Nothing runs in isolation — every step feeds the next, and the customer never sees the seams.
Every inbound channel now feeds into a single pipeline. AI reads the inquiry, extracts the essential data (vehicle type, origin, destination, timeframe), filters obvious spam or tire-kickers, and pushes real leads into the CRM within seconds. The team sees a clean, ranked list of what actually needs their attention.
When a quote is accepted, the system checks driver availability, matches by region and vehicle type, and proposes an assignment. One click confirms it. The driver gets the details, the customer gets the pickup window, and the schedule updates everywhere at once.
At pickup and delivery, the driver fills out a structured inspection form on their phone. AI validates the inputs against the vehicle history and photos, flags anything inconsistent, and generates a consistent PDF report. No more "I forgot to check the rear bumper" disputes three weeks later.
The most important choices weren't about which tools to use. They were about where to draw the line between what the AI handles and what a human decides.
The AI drafts customer replies, driver messages, and inspection summaries — but nothing goes out until a human clicks approve. For the first two weeks, that approval was mandatory on every message. As the drafts proved consistently accurate, we gradually widened the window. Even today, customer-facing pricing always requires a human yes.
Every AI step is forced to output a strict JSON structure. It can't ramble, can't invent fields, can't "helpfully" add extra data. If the input doesn't match the schema, the workflow routes the message to a human instead of guessing.
The whole system runs on their own cloud account, in their own jurisdiction. We have access during the deployment and stabilization phases. After 90 days, we revoke our own credentials. If they ever leave us, they keep every piece of the system — no exports, no migrations, no friction.
We could have built this faster with no-code connectors. We didn't. Every integration has explicit error handling, retry logic, and logging. When a third-party API goes down (and it will), the workflow doesn't lose data — it queues it and picks back up when the service recovers.
The 90-Day Protocol isn't a slogan. Here's what it actually looked like for AutoTransfer.
We watched every transaction live. No autonomous action — the system processed messages but didn't send anything without our review first. Two edge cases surfaced (a non-standard vehicle description and a WhatsApp message in mixed Czech/German) which we handled manually and added to the schema validation rules.
Every outbound message routed to Slack for one-click approval. The team approved about 200 messages in that window. By day 10, they were approving without reading most of them — a sign the AI drafts were reliably accurate.
Customer-facing responses went autonomous. Pricing confirmations still required a human. We used the stabilization window to tighten the inspection protocol templates and add automatic dispute-prevention (photo timestamps, GPS coordinates on inspection forms).
We transitioned the system to run without our intervention. Monitoring stays on, but the daily operation is theirs. Their team now handles about a third of the admin load they used to.
"We used to drown in manual data entry and phone calls. Now the system handles the busywork, and we focus on driving and customer care."
AutoTransfer
Vehicle Logistics · CZ/EU
The team saves roughly 25 hours a month on admin work — the equivalent of adding a part-time employee without hiring one. More importantly, they're converting more leads because response time no longer costs them the sale.
The system is under active optimization. Recent additions include an AI-assisted quote generator for common routes and automatic multi-vehicle batch handling. Both came out of the monthly ROI reviews — the ones that keep finding new bottlenecks to eliminate.
Every new feature is built on the same infrastructure. Nothing requires a new contract or a new vendor. The system grows as the business does.
Every engagement starts with the same free call. Bring your bottleneck — we'll tell you honestly if automation fits.
Step 1: Free Discovery Call
You explain the operation. We explain how we'd approach it.
Step 2: The Diagnostic Audit
If it's a fit, we map your workflows in detail and deliver a concrete implementation blueprint.
20 minutes. No obligation. We'll tell you honestly if we can help.