Years ago, I worked in the refining industry. Crude came in, the distillation columns ran, and different product trains ran off the side, and all of it reported to one control room. A wall of screens showed every critical process: pressure up, temperature down, a flow that stopped moving. The operators didn’t wander the plant hoping to find problems. The problems came to them, ranked by what mattered, and they acted.
Freight has never had that control room that overseas all company-wide processes – operations, accounting and financial. In most freight companies, the problems sit inside a dozen screens until somebody goes digging. A load with a pickup coming and no carrier on it. A rate confirmation that went out and never came back signed. A driver whose ELD went dark an hour ago. A load running a negative margin that nobody has noticed yet. Somebody eventually finds each one, and the finding takes longer than the fixing.
That’s the gap EKA Control Center AI, one of the four new agents we just announced, was built to close.
What It Actually Does
Control Center monitors at least 25 signals across an operation and pushes up the ones that need a person, on a single screen, sorted by severity. The signals come in two kinds. Some are data problems: paperwork missing after a delivery, an asset document about to expire, a geocode that doesn’t check out. Some are event problems, where a threshold you set gets crossed: a pickup clock running short, detention building at a dock, a delivery window at risk, a margin going negative.
The number matters less than what’s behind it. We started with these 25 because we run a TMS real-time, all day and know where operations actually break. Depending on the freight you run, the list can grow toward 50, and every signal traces to a real failure we’ve watched cost somebody money.
It’s also built around how a team actually divides work. An admin sees the whole operation. An ops person sees the loads and business units they own, and nothing else, so nobody sifts through another desk’s problems. Each person picks the signals they want in their own settings. If geocodes are someone else’s job, you can snooze the task. If a teammate is already on an item, snooze it, and it drops off your board for 12 hours. The board refreshes near real-time, and you can have it open as your first screen of the day.
Start the Day From a List Instead of a Hunt
Here’s the workflow in practice. You log in, and the day is already triaged: highest-severity items at the top, each with the load, details, and contacts attached. The negative-margin load needs you, so you click through, see the problem, assign the right carrier, and send the rate confirmation without leaving the flow. The item drops off. The next one is someone else’s, so it gets snoozed. Ten minutes in, you know exactly what your day is.
It works like an inbox you run to zero. The job stops being hunt-for-problems and becomes judge-and-fix, which is the shift we wrote about in moving people from pushing screens to making decisions. The math gets better as you grow too: more loads mean more exceptions, and every exception found near real-time instead of hours is margin kept. The win isn’t only the time your people save. It’s the cycle time on the exception itself, because a problem caught at 8:15 costs less than the same problem found at 2:00. It results, among other things, in employee empowerment.
Alerts First. Then Analysis. Then Action.
We’re deliberately starting simple. Step one is what ships now: surface the exception in real-time to matter, with intelligent workflow to quickly identify the root cause of the problem and then to quickly to fix it. No razzmatazz. Step two is intelligent analysis at your fingertips, where the system helps answer why: what’s behind the miss, what the options are, and which one looks best. Step three is action, where routine cases are resolved, and a person handles only what requires judgment. AI-driven solutions are planned for 2027 to autonomously fix many of the exceptions.
Each of those 25 signals can climb that ladder on its own schedule. A rules-based fix, such as assigning a known carrier to a known lane, can quickly become a semi-autonomous or autonomous agent. The heavier problems take longer, and that’s fine. We’d rather move one step at a time with customers than hand anyone a mouthful they can’t swallow. By 2027, plenty of these signals will run autonomously for the operations that want them to. The ones that don’t can use the intelligent workflow identify the root cause of the problem, fix it and still be far ahead of digging through screens.
There’s a longer game in the data too. Once you can see all your exceptions in one place, you can finally ask why you have so many. Which ones are preventable? Which customer, lane, or process keeps generating them? Fix those at the source, and the board gets quieter every month. Fewer exceptions is the actual goal. The control room is how you get there.
Why This Needs One Environment
None of this works as a bolt-on. A control center is only as good as the data feeding it, and ours reads the operation’s own record: the loads, the clocks, the documents, the margins, all living in the same Omni-TMS environment where the work happens. Stitch the same idea across a stack of disconnected tools, and you inherit every seam between them, and the alert arrives late, wrong, or not at all.
Native also means the fix is one click away instead of three systems away, and it means the signals share context. Detention building at a dock affects the delivery window at risk because DockTime and On-Time use the same record.
A quick story about what that foundation makes possible. One of our broker customers recently started taking overflow freight from a large retail shipper. The orders arrive via EDI, are rated in the TMS, land in active loads, and are assigned to the carrier without a single person touching the sequence. Nobody rekeyed anything. That sequence runs on the environment underneath Control Center, and it’s the same reason the control room can exist at all.
The Bottom Line
Every refinery, every power plant, every operation where minutes are money runs from a control room. Freight qualifies. EKA Control Center AI gives an operation that room: about 25 signals, ranked by severity, scoped to each role, one click from the fix, and a path from alerts to analysis to action as fast or as slow as your team wants to climb it. It’s part of the platform, not another tool on the pile. Talk to EKA, and we’ll show you your first freight Control Center.
FAQs
What is EKA Control Center AI?
A control room for a freight operation, built into EKA Omni-TMS. It monitors about 25 operational signals, including data problems like missing paperwork or expiring documents, and event problems like short pickup clocks, detention, or negative margins, and pushes the ones that need a person onto a single screen, sorted by severity, with a direct path to the fix.
How is this different from the alerts my TMS already sends?
Most TMS alerts are scattered notifications with no ranking, ownership, or workflow. Control Center is one board per person, scoped to their role and business units, sorted by cost (highest first), with snooze and settings so each user can tune their own view. The alert, the context, and the fix live in the same place, so the cycle time from problem to resolution drops.
Is this going to automate people out of their jobs?
No. It takes away the hunting, and hunting was never the job. The path runs in steps: first, surface the exception; then, help analyze it; then, automate only the routine cases, with a person handling everything that requires judgment. Teams move up that ladder at their own pace, one signal at a time. The people we build for end up doing more of the work that matters, and less of the digging.
Do I have to use all 25 signals?
No. Each user can select the signals relevant to their role and snooze items a teammate is handling. An admin sees everything, an ops user sees their own loads, and the list itself can grow to 50 signals depending on the freight you run. It’s your control room, so you decide what’s on the wall.