By Eric Rempel, Chief Innovation Officer, Redwood Logistics
Everyone is talking about AI automation in supply chain. AI will reroute freight, schedule appointments, handle check calls, answer inbound calls, communicate with carriers, resolve invoice issues, manage exceptions, and recommend the next best action. All true, and all enormously valuable.
Across logistics, there is still a surprising amount of human effort sitting between highly automated systems. A transportation management system may automatically build and tender a load, only for a person to spend the next several hours calling carriers, moving an appointment, answering a customer email, resolving a document problem, or figuring out what to do when reality no longer matches the plan.
AI agents are going to absorb a large amount of that work. Voice agents will answer calls, and other agents will place them. Agents will communicate by text and email, work inside transportation processes, interpret documents, negotiate routine issues, and coordinate across parties. Eventually, the companies on both sides of a transaction will have agents, and agent-to-agent communication will become normal.
That is going to create a massive productivity unlock. But I think it also raises a much more interesting question:
What is the future of supply chain after AI automation?
Assume the Automation Works
This industry is not starting from zero. It is already one of the most automated operating environments. Orders flow between systems, loads are built, rates are calculated, routing guides execute, carriers are tendered, EDI and APIs move information between companies, status events are captured, invoices are matched, and exceptions are surfaced. In a highly automated shipper or managed transportation operation, the vast majority of transactions may already flow without direct human intervention.
Humans tend to live in the gaps. A carrier rejects. An appointment changes. A customer makes a new request. A facility is constrained. A document is wrong. The system cannot understand the context, and the plan no longer matches reality.
For the last 20 years, the industry's answer has largely been "manage by exception": let the system handle what it can, watch for variance, alert a human, and ask the human to intervene. AI is going to compress that operating model.
So let's assume it works. Assume AI eventually handles 99% of today's repetitive communication, coordination, and manage-by-exception workflow. Then what?
The Unit of Human Work Gets Larger
I do not believe the answer is simply fewer people doing less work. I think the altitude of human work changes.
Today, someone may be responsible for one shipment, one carrier call, one appointment problem, or one customer escalation. Tomorrow, an agent may handle thousands of those interactions, and the human no longer needs to sit inside every conversation. Instead of negotiating one load, someone can design a better capacity strategy. Instead of repeatedly calling carriers, someone can determine which carrier relationships the company should deepen and why. Instead of answering one customer issue at a time, someone can ask what the entire customer experience should feel like. Instead of manually managing one workflow, someone can orchestrate a system of people, platforms, partners, and intelligent agents around a larger outcome.
The work moves from the shipment to the relationship, from the relationship to the network, from the network to the customer experience, and eventually from the customer experience to the business model itself. AI does not just eliminate work.
The Shipment Gets Automated. The Human Inherits the System.
Consider a highly automated transportation operation. The TMS builds the shipment, the routing guide identifies the carrier, the tender is sent electronically, and tracking information flows back into the system. Today, a lot of the remaining human work begins when that clean flow breaks: the carrier rejects the tender, the facility cannot make the appointment work, the shipment is running behind, or the customer changes a requirement. The transportation team steps in.
AI will increasingly handle those gaps. The mistake would be to assume the transportation team should simply wait around for a more difficult class of exception. The work should move up. Why does this lane keep failing? Why are we buying transportation this way? Why does this facility consistently create downstream cost? Why does a customer promise conflict with the physical capabilities of the network? Why are we treating a recurring problem as an exception? Could two customers' networks create an opportunity together? Could a different service model create growth?
Once the transaction is automated, the human inherits the system around the transaction. That is a much bigger job.
Customer Expectations Will Move Up Too
This shift will not happen only inside logistics operations. Customer expectations are going to change.
Today, we often define great service as responsiveness. You call me, I answer. You email me, I respond. You have a problem, I fix it.
But once people live in a world of AI systems that remember context, the questions will change. Why did I have to explain this again? Why didn't you remember that this customer prioritizes service over cost? Why didn't you recognize that this facility behaves differently at the end of the week? Why are you giving me a generic recommendation when you have years of history with my business? And perhaps most importantly: why did I have to ask?
The bar moves from responsive to anticipatory, and that requires a much deeper understanding of the customer. Not just their shipments. Their preferences, their customers, their priorities, their risk tolerance, their promises, their operating patterns, and their definition of value all change. The logistics companies that understand a customer's operating DNA will be able to orchestrate far more personalized, flexible, and differentiated solutions. This is not just better customer service. It is an entirely different customer relationship.
Where Does Continuous Improvement Come From?
This may be the question I find most interesting. If AI handles repetitive work, where do strategy and innovation come from?
Today, continuous improvement is often episodic. We hold a quarterly business review, run an annual procurement event, launch a network study, bring in a consulting team, or analyze a problem after it has become painful enough to attract attention. That model exists partly because people are busy. They are running the operation, resolving exceptions, answering customers, and keeping the machine moving.
But imagine an operating environment that continuously observes the shipments, carrier behavior, service outcomes, facility performance, customer conversations, cost, and the decisions people make. Improvement no longer has to wait for an annual redesign cycle. The company brain can continuously surface questions. Why does this keep happening? Why is this policy still in place? Why did our best operator make a different decision here? Why does one customer respond differently than another? Where is friction building? Which constraint is real, and which one have we simply accepted?
The AI may find the pattern, but humans still have to decide what matters. They need to connect dots across domains, challenge assumptions, understand relationships, set ambitious goals, imagine alternatives, and decide what kind of company they are trying to build. That is where I believe human judgment becomes more important, not less.
AI gives people more signal and more time. People exercise judgment. AI learns from those decisions and outcomes. The company gets smarter, and the cycle repeats. The future of continuous improvement may not be a quarterly project. It may be a continuous learning loop between the human brain and the company brain.
AI Plus Corporate DNA
This leads to an important question about competitive advantage. If every company has access to increasingly powerful general AI models, what makes one company's AI better than another's?
The model alone will not be enough. A shipper and its competitor may use the same model. Two logistics companies may use the same model. But they do not share the same history, the same customers, the same partner relationships, the same service recoveries, the same operating decisions, the same judgment calls, the same culture, the same definition of what a great customer experience looks like, or the same way of winning.
Today, much of that knowledge is trapped in people. A great operator knows that one facility behaves differently after 2 p.m. A customer leader understands that a particular shipper may talk about cost but makes decisions based on service risk. A transportation expert knows that a carrier may look weak in the aggregate but is exceptional in a particular operating environment. Traditional systems capture the transaction, the timestamp, and the reason code. They rarely capture the judgment.
AI working side by side with a company's people has the potential to change that. Over time, it can begin to understand how a company operates, makes tradeoffs, protects relationships, manages risk, and creates value. General AI will give general answers. AI shaped by a company's people, history, workflows, and decisions can create specialized advantage. I think of this as AI plus corporate DNA.
The real question for companies is whether that intelligence becomes part of the company's institutional memory or disappears every time a great employee walks out the door.
From Execution to Imagination
If we extrapolate further, the most interesting future may not be a supply chain that simply runs itself. It may be a supply chain that helps the company imagine what is possible.
Today, supply chain largely fulfills promises the business has already made. Sales sold it, the customer ordered it, and we out how to deliver it. But imagine a supply chain continuously learning across customer behavior, inventory, facilities, carrier capacity, cost, service, and risk. It may start revealing entirely new possibilities. We could serve this customer differently. We could make a different promise in this market. We could combine these flows, create a new service, change where inventory is positioned, or enter a market because a constraint we assumed was fixed is no longer fixed. We could design the customer relationship around what we now understand about their business.
At that point, supply chain is no longer only executing business strategy. It is helping create business strategy. As AI absorbs the transaction, humans move to the relationship. As AI learns the relationship, humans move to the network. As AI understands the network, humans get to imagine what the business can become. Post-automation supply chain is not just adaptive. It becomes imaginative.
The Orchestration Challenge
There is one problem. This future will not exist inside a single system.
No TMS owns the entire supply chain. No brokerage platform does. No ERP, visibility provider, warehouse system, carrier platform, or AI model does either. The supply chain is already an ecosystem of systems, companies, partners, and people, and now we are about to add intelligent agents to that environment: a shipper agent, a transportation agent, a logistics provider agent, a carrier agent, a facility agent, a customer agent. They will communicate, make decisions, share approved context, react to change, escalate, and learn.
The future requires more orchestration, not less. Someone has to connect the operating model. Someone has to think across strategy, execution, technology, partners, and intelligence. Companies need a foundational fabric where systems and agents can interact safely, where context can travel, where decisions can be governed, and where learning can compound. This is a major part of the thesis behind Redwood's Open 4PL model.
The future logistics provider cannot simply sell capacity, manage transportation, or implement another piece of software. The role has to become broader: help companies design the strategy, deliver the operating model, orchestrate the ecosystem, continuously improve it, and provide the technology fabric underneath it that allows the entire environment to evolve.
Because the destination is not an autonomous shipment. It is a learning enterprise. The companies that win will combine the judgment of their people with the memory of their company. They will use AI to absorb low-judgment work, but they will also create an environment where human expertise is amplified, captured, and continuously pushed into larger problems. Work will move up: from the shipment, to the relationship, to the network, to the business.
And that raises the next question I think supply chain leaders need to confront:
What does the org chart look like to manage something like this?