In August one of our own services stood still for a week without anyone noticing. It started every day, reported "running" and delivered nothing, because the step before it had died on a network dropout. The watchdog checked the run, not the result. Since that day all our watchdogs check the result.
That is why we run our own company with the same services we build for forwarders. About 80 automated processes, 14 machines, one person. Not because that is especially elegant, but because every mistake happens at our place first.
The 80 processes are not robots making coffee. They are services like the ones at our customers, except the customer is us:
About 400,000 records a month pass through these services. The number matters less than the fact that we can look up every one of them: which run, which step, which result.
Honesty is part of the positioning, so three things that did not work:
The watchdog that checked the run. See above. A week of standstill, visible only in hindsight. The rule from it now sits in every customer offer: the alarm hangs on the result. Zero bookings on a working day with input is an outage.
The agent that decided when in doubt. In an early version a service was allowed to choose on its own in unclear cases. It chose plausibly and wrongly. Since then: when in doubt, review list, with a reason, and a person decides. That costs minutes and saves weeks.
The reconciliation that was too generous. A check service compared supplier names fuzzily and let a payment through because two companies carried the same word in their name. Since then we test thresholds with real cases, not examples, and a match above the threshold is a review case, not a release.
Every service for a forwarder inherits three things from our own operation:
What does not carry over: our choice of tools. Which model, which database, which browser automation we use for ourselves depends on our operation. At the customer it depends on his, on his systems and on who operates the service in two years.
Strategy& and BVL surveyed 115 logistics providers in November 2025: 96 percent have started digitalization, 10 percent scale. Scaling means services run without supervision and somebody notices when they do not. That is not a question of the model. It is a question of protocol, watchdog and review list, and we learned those on ourselves before we asked a customer to live with them.
A caveat: our 80 processes are smaller than those of a forwarder with 200 trucks. What has two runs a day at our place has 20 there. The rules stay, the thresholds in the watchdog change.
What would you look up first in our protocol? Tell me which service interests you. I will show you one run, with everything that went wrong. Ask for a process check: 30 minutes, no sales pitch.
