Latest / Logistics at a Crossroads / S2 🎙️Episode 15 — AI Can Optimize the Route. Can It Run the Operation?
Transcript
- 1:14SPEAKER_00: Hey, hey, hey, and welcome back to Holding the Line.
- 1:17SPEAKER_00: It's your girl, Gia.
- 1:20SPEAKER_00: And I gotta tell you, everybody everywhere has an AI announcement right now.
- 1:26SPEAKER_00: A new AI dashboard, an AI co-pilot, an AI assistant that supposedly is about to make the whole supply chain so smooth that nothing is ever late again.
- 1:41SPEAKER_00: Oh wait, cheer.
- 1:43SPEAKER_00: And I want a doc appointment that means what it says.
- 1:47SPEAKER_00: But no, really.
- 1:48SPEAKER_00: Here's the actual headline.
- 1:50SPEAKER_00: In the 2026 State of Logistics report, released June 16th, AI had moved from something companies were trying to something delivering results in specific, well-defined supply chain jobs.
- 2:07SPEAKER_00: But adoption was still kind of uneven.
- 2:11SPEAKER_00: Because there is a big difference between a tool that helps one, helps on one workflow, and an AI that is truly inside the core operation.
- 2:24SPEAKER_00: And that is the part I want us to sit with.
- 2:28SPEAKER_00: Because the question has never been can AI optimize a route?
- 2:33SPEAKER_00: Of course it can.
- 2:34SPEAKER_00: It can definitely help.
- 2:37SPEAKER_00: I mean, AI can cut down, sort through more shipment shipment signals than one person can reasonably even think about or chase down before lunch.
- 2:51SPEAKER_00: I mean, AI has the ability to spot an estimated time of arrival arrival, aka the ETA, that is drifting, or group similar exceptions, forecast demand, suggest a carrier, and put the problem closer to the top of the screen.
- 3:12SPEAKER_00: That is useful.
- 3:15SPEAKER_00: That is great AI.
- 3:19SPEAKER_00: But can it run the operation?
- 3:23SPEAKER_00: Well, operations have a terrible habit of becoming Tuesday.
- 3:29SPEAKER_00: Tuesday is somebody called out.
- 3:31SPEAKER_00: A crane went down, a vessel changed berths, a customer changed priority, containers on hold, traffic back up, drivers early, drivers late, rain stopped, the whole shebang.
- 3:46SPEAKER_00: Couldn't see nothing.
- 3:48SPEAKER_00: Ship stopped, wind.
- 3:51SPEAKER_00: Um, and our system might say that equipment is available when the person standing outside says it is absolutely not.
- 3:59SPEAKER_00: Hi, this is logistics.
- 4:01SPEAKER_00: Welcome.
- 4:02SPEAKER_00: We've been waiting for you.
- 4:04SPEAKER_00: Now I I know I made that kind of light, but that's truly what a Tuesday looks like.
- 4:11SPEAKER_00: And with AI, it has the ability to see a pattern in those signals.
- 4:18SPEAKER_00: It may tell you that the 12 loads are likely to miss delivery.
- 4:22SPEAKER_00: Awesome.
- 4:23SPEAKER_00: That gets me the planner out of detective mode.
- 4:27SPEAKER_00: But then comes the groom folk part.
- 4:31SPEAKER_00: Which customer gets the remaining capacity?
- 4:34SPEAKER_00: Which freight can actually wait?
- 4:37SPEAKER_00: Is my inventory right?
- 4:39SPEAKER_00: Or is somebody looking at a ghost?
- 4:42SPEAKER_00: Who calls the customer?
- 4:44SPEAKER_00: Um, who is authorized to spend the money?
- 4:47SPEAKER_00: I'm not.
- 4:47SPEAKER_00: Somebody else needs to make these decisions, not just the calculations.
- 4:54SPEAKER_00: And the decision does not become less owned because a recommendation arrived in a very confident font.
- 5:01SPEAKER_00: Look, data matters.
- 5:04SPEAKER_00: AI does not sprinkle fairy dust on a bad handoff.
- 5:09SPEAKER_00: If the appointment feed is stale, the inventory status is wrong, or carrier updates come late, like the basic reference data, locations, customers, products, equipment, does not match across partners, the recommendation can be polished nonsense.
- 5:31SPEAKER_00: And sure, we can absolutely optimize a fantasy.
- 5:35SPEAKER_00: Congratulations.
- 5:37SPEAKER_00: We just digitize a mess.
- 5:41SPEAKER_00: Stop.
- 5:41SPEAKER_00: Let me read back.
- 5:43SPEAKER_00: Let me go back.
- 5:44SPEAKER_00: That's not a reason to throw out the AI tools.
- 5:48SPEAKER_00: It is a reason to use them like adults.
- 5:52SPEAKER_00: And I mean that.
- 5:55SPEAKER_00: So right now in transportation and logistics, the best use case may be letting the system surface the exception.
- 6:06SPEAKER_00: Let the AI go in, gather relevant facts, and show options.
- 6:11SPEAKER_00: Because that allows the planner to spend less time clicking through five screens and more time deciding what to do.
- 6:20SPEAKER_00: And that is not human versus AI.
- 6:24SPEAKER_00: That is what we call human plus AI.
- 6:29SPEAKER_00: With the human, the planner still responsible for the call.
- 6:36SPEAKER_00: And come on, let's be honest about what it takes.
- 6:40SPEAKER_00: People need training, context, practice, support, and a clear path to say this does not match how work actually happens.
- 6:52SPEAKER_00: And that feedback is gold.
- 6:56SPEAKER_00: It is not a resistance, no matter what your manager just looked at you and said.
- 7:51SPEAKER_00: Stop and listen.
- 7:56SPEAKER_00: Because the person giving you this feedback, they're trying to be feed forward because of their experience on that terminal.
- 8:10SPEAKER_00: Seeing what happens, knowing two trucks or one truck can't be in two places at one time.
- 8:18SPEAKER_00: Look, a good planner may not be able to give you the fancy equation for why a load needs attention.
- 8:25SPEAKER_00: Sometimes the answer is I've seen this before.
- 8:30SPEAKER_00: The ship's update changed, the drayage pulls tight, the the consonee has a narrow receiving window.
- 8:38SPEAKER_00: And suddenly that tidy route is not so tiny.
- 8:42SPEAKER_00: That's not mystical.
- 8:44SPEAKER_00: It is pattern recognition built through real work.
- 8:50SPEAKER_00: And maybe a few little scars from being in the trenches with it and being the one having to dig everybody out.
- 9:01SPEAKER_00: AI can build pattern recognition too.
- 9:06SPEAKER_00: And in some narrow jobs, it may see a risk sooner or handle the volume better than any one person can.
- 9:17SPEAKER_00: Let it do it.
- 9:24SPEAKER_00: Planners should not have to spend three hours hunting for an exception that a decent system can find in three seconds.
- 9:33SPEAKER_00: Let AI do the work.
- 9:37SPEAKER_00: Let them come in.
- 9:39SPEAKER_00: Show me.
- 9:48SPEAKER_00: But you knew it was coming.
- 9:52SPEAKER_00: You turn a recommendation into an operating rule.
- 9:57SPEAKER_00: Can we stop and answer a few of your planner questions?
- 10:02SPEAKER_00: And here they are.
- 10:04SPEAKER_00: What decision is the system allowed to make on its own?
- 10:08SPEAKER_00: And what decision does it need a person for?
- 10:12SPEAKER_00: Who can override it?
- 10:15SPEAKER_00: And if someone overrides it, what should that trigger do?
- 10:22SPEAKER_00: Who owns the outcome if the recommendation the AI gave us is wrong?
- 10:29SPEAKER_00: How are we verifying or checking the inputs, the recommendations, and the actual results after the fact?
- 10:40SPEAKER_00: And if one of you come to me and say the computer said so, I swear to the heavens, you and I are going to go outside the gate.
- 10:49SPEAKER_00: Because all that does is that's how we've always done it.
- 10:53SPEAKER_00: No, no, no.
- 10:55SPEAKER_00: Those are lazy answers.
- 10:58SPEAKER_00: One hides behind a machine and one hides behind a bad habit.
- 11:05SPEAKER_00: And neither one of those tell the customer or the person asking the questions why something is late, and it surely does not help the dispatcher decide what happens next.
- 11:17SPEAKER_00: Okay, leaders, listen up.
- 11:21SPEAKER_00: Because you have a job here too.
- 11:25unknown: Okay.
- 11:25SPEAKER_00: I say this with love.
- 11:28SPEAKER_00: Do not buy a shiny tool, hand out logins, and call that transformation.
- 11:36SPEAKER_00: No.
- 11:37SPEAKER_00: Name an accountable owner.
- 11:39SPEAKER_00: Bring operations, IT, procurement, and the people who are to work the exceptions into the room before rollout.
- 11:53SPEAKER_00: Not after you've decided to roll it out or during the rollout.
- 12:00SPEAKER_00: We need to define the decisions, test the cases, and keep an incident playbook for when the tool is wrong.
- 12:12SPEAKER_00: The feed goes down, or the operation changes faster than the model does.
- 12:19SPEAKER_00: Because a tool that can predict trouble is helpful.
- 12:23SPEAKER_00: I'm not gonna lie.
- 12:27SPEAKER_00: A team that knows who decides, who communicates, and who can change course when trouble arrives, baby, that's what holds the line.
- 12:39SPEAKER_00: And that might take us into the next conversation.
- 12:42SPEAKER_00: Look, when I said bring AI in, let them run the forecast, let them run the routes, let them do things, let them recommend the move.
- 12:53SPEAKER_00: But automation is changing the physical operation too.
- 12:58SPEAKER_00: Because robots can move, sort, scan, and retrieve freight.
- 13:02SPEAKER_00: Fine.
- 13:04SPEAKER_00: They can with probably less accidents.
- 13:07SPEAKER_00: But you know what you still need?
- 13:09SPEAKER_00: People.
- 13:10SPEAKER_00: They have to operate the system, maintain it, question it, and recover the work when it stops behaving like the sales demo.
- 13:21SPEAKER_00: Remember that.
- 13:23SPEAKER_00: Shiny and new is great.
- 13:25SPEAKER_00: Bring the people that have to work with it, that know what's going on today before you roll it out.
- 13:33SPEAKER_00: So here's my question to you.
- 13:35SPEAKER_00: Because we all know season two is all about questions.
- 13:38SPEAKER_00: If AI gave your operation the perfect recommendation tomorrow, would your people know when not to follow it?
- 13:49SPEAKER_00: And who has the authority to make that call?
- 13:55SPEAKER_00: Exactly.
- 13:57SPEAKER_00: Thank you for tuning in to holding the line.
- 14:00SPEAKER_00: I'm your host, Gia.
- 14:02SPEAKER_00: And until next time, I'll be right here, navigating the crossroads with you.