At 5:40 Monday morning, my new boss dropped forty-three computer-generated delivery routes onto my desk and told me to approve them before the drivers arrived. I had managed dispatch operations at Great Lakes Freight outside Cleveland for fourteen years. Marcus Bell had been operations director for twelve days.
I opened the routing dashboard. The software promised nine percent lower fuel costs and nearly six fewer driving hours across the fleet. Marcus stood behind me, pleased with himself. “One click, Dana. Approve all.”
I didn’t click.
Route 17 immediately bothered me. It sent a fully loaded tractor-trailer carrying nearly 78,000 pounds through Mill Creek Road, a narrow county route I knew well. I zoomed in and saw the system had chosen it to avoid eleven minutes of interstate traffic.
“Mill Creek has restrictions,” I said.
Marcus sighed. “The software accounts for restrictions.”
“Not all of them.”
His expression hardened. He had spent his first two weeks telling everyone our experienced dispatch team relied too heavily on “tribal knowledge.” Now he leaned over my desk and said, “Are you afraid technology might finally make your job unnecessary?”
I ignored the insult and checked more routes. Three sent long trailers toward tight industrial crossings. Another directed a truck under an overpass whose clearance data had recently changed. Route 31 showed a delivery entrance that had been permanently closed.
I refused to approve all forty-three.
Marcus raised his voice loud enough for the dispatch floor to hear. “If you cannot perform a basic responsibility because a computer makes you uncomfortable, maybe we need someone who can.”
That was when veteran driver Frank Donnelly walked in. Frank had driven commercial trucks for twenty-seven years. Marcus shoved the printed Route 17 toward him. “Tell Dana whether you can handle this.”
Frank studied it silently.
Then his face changed.
He pointed at Mill Creek Road. “You want me taking the loaded rig here?”
“The optimization system selected it,” Marcus said.
Frank looked around the room, then back at Marcus. “Okay. One question.”
Nobody moved.
“Which one of us is calling the state police when that trailer reaches the weight-restricted bridge?”
Marcus stopped speaking.
Frank tapped the map again. “Because if you want me to drive seventy-eight thousand pounds across a bridge posted for less, I want your order in writing before I turn the key.”
The dispatch room went completely silent.
I rotated my monitor toward Marcus.
“Still want me to click approve?”
Marcus did not answer. Instead, he grabbed the route sheet from Frank and stared at it as if the paper itself had betrayed him. Then he said the software vendor’s database must contain more current information than a roadside sign Frank remembered seeing.
Frank pulled out his phone.
He had photographed the bridge sign four days earlier because road crews had changed the restriction after an inspection. The new limit had not yet appeared in the commercial routing database feeding our system.
Marcus’s face lost some color.
I asked everyone to hold departures for twenty minutes while we reviewed the routes manually. Marcus objected until our safety manager, Luis Ortega, walked in and heard what had happened. Luis immediately supported the delay.
We divided the forty-three routes among dispatchers and senior drivers. Within fifteen minutes, the screen that had promised perfect efficiency began producing problems. Route 8 used a road temporarily closed for construction. Route 14 included an impossible left turn for a fifty-three-foot trailer.
Route 22 was worse. The system directed a refrigerated truck through a residential street where commercial vehicles over a posted weight were prohibited. Route 31 sent another driver to the abandoned entrance of a distribution center, leaving no safe space for a tractor-trailer to turn around.
Nobody argued that the software was useless. It was actually excellent at comparing mileage, fuel consumption, appointment windows, and traffic patterns. The problem was Marcus had treated optimization as authorization.
Luis documented every error.
By eight o’clock, we had corrected nine routes. The remaining thirty-four were approved after verification. Departures were delayed, but every truck left with a route a dispatcher and driver had reviewed.
Marcus disappeared into his office.
That afternoon, he called me in with HR present. For a moment, I thought he intended to fire me anyway. Instead, HR asked me to explain exactly why I had refused his instruction.
I showed them my notes.
I had recorded the time Marcus requested blanket approval, the specific routes I questioned, the reasons for each concern, and the corrections made afterward. I also showed them an email I had sent three days earlier recommending that computer-generated routes require human verification during the system’s rollout period.
Marcus had replied with two words.
Not necessary.
The HR manager read that message twice.
Then she looked at Marcus.
“When Dana raised a safety concern this morning,” she asked, “why did you threaten her employment instead of verifying the route?”
For the second time that day, Marcus Bell had no answer.
The company’s regional vice president arrived two days later. By then, the routing vendor had confirmed what we already knew: its software depended on multiple third-party data sources, and temporary restrictions could appear after database updates. The system had never been marketed as a replacement for professional driver judgment.
That detail mattered.
Marcus had told senior management that automated routing would allow the company to reduce dispatch review time by nearly eighty percent. He had presented the system as something that could make decisions independently. The vendor’s implementation documents said the opposite.
The regional vice president asked Frank, Luis, Marcus, and me to attend a meeting.
Frank brought the photograph of the bridge sign.
I brought the forty-three original routes and our corrected versions. Luis brought the safety report. Marcus brought a presentation explaining that employees had resisted technological change.
The vice president stopped him halfway through.
“This isn’t about whether people like technology,” she said. “It’s about whether we use it correctly.”
She asked me what I recommended.
I proposed a simple process. The software would generate initial routes because it could analyze thousands of variables faster than any dispatcher. Dispatchers would verify restrictions, construction alerts, delivery entrances, and unusual road conditions. Drivers could flag anything inconsistent with what they saw in the field.
Nobody would blindly trust either a computer or a person.
The company adopted the procedure that week.
Marcus kept his title, but responsibility for transportation safety and route approval was removed from his direct control during the remainder of the rollout. He was also required to complete additional compliance and fleet-safety training.
A month later, the routing system was saving us more fuel than before. Once drivers realized reporting errors improved the database instead of being treated as resistance, they started participating. The number of route corrections dropped steadily.
Marcus rarely spoke to me unless necessary.
Frank, however, developed a habit of stopping beside my desk every Monday morning. He would place his coffee down, glance dramatically at the routing screen, and ask, “Computer trying to kill me today?”
I always laughed.
Six months later, I was promoted to regional dispatch and systems coordinator. My job was not to fight automation. It was to make sure the people using it understood what automation could and could not know.
On my first day in the new position, Frank handed me a small framed photograph.
It showed the Mill Creek bridge sign.
Underneath, he had written one sentence:
Eleven minutes saved. One disaster avoided.
I kept it beside my monitor.
Because the smartest decision I made that morning wasn’t rejecting technology.
It was refusing to stop thinking.
