Building Intelligent Shipping Operations Beyond Automation: Perspectives from Kyle Henzel, COO of Ship.com

What's Inside: AI is reshaping logistics, but the biggest opportunities are no longer in experimentation. They are in execution. In this conversation, Kyle Henzel, President and COO of Ship.com speaks to Sarrah Pitaliya, VP of Digital Marketing at Radixweb and shares where AI is delivering measurable value today, what logistics leaders still get wrong about automation, and how shipping organizations can prepare for increasingly intelligent and autonomous operations.

The logistics industry has always been a data-intensive business. Every shipment, route, warehouse movement, customer interaction, and delivery event generates information that can be used to improve performance. What has changed over the last few years is the industry's ability to act on that information in real time.

AI is rapidly moving from isolated optimization projects into core logistics operations. Organizations are using it to improve route planning, predict disruptions, automate repetitive workflows, enhance customer experiences, and create more resilient supply chains. At the same time, rising customer expectations, growing operational complexity, labor shortages, and margin pressures are forcing logistics providers to rethink how technology supports day-to-day execution. Industry research shows that AI investment across logistics continues to accelerate, but many organizations still struggle with data quality, system integration, and operational adoption.

At Radixweb, we work with organizations building AI-powered software systems that turn operational data into actionable business outcomes. To better understand how AI is transforming logistics from the inside, Sarrah Pitaliya, our VP of Digital Marketing, spoke with Kyle Henzel, President and COO of Ship.com.

Drawing from his experience helping ecommerce sellers, direct sales consultants, and fulfillment-driven businesses simplify shipping operations, Kyle shares practical perspectives on AI adoption, automation, operational bottlenecks, and the future of intelligent logistics platforms.

About Kyle Henzel

Kyle Henzel is the President and Chief Operating Officer of Ship.com, a SaaS shipping platform that helps eCommerce brands simplify fulfillment and protect margins. With extensive experience in logistics and eCommerce operations, Kyle leads shipping rates, carrier strategy, and platform integrations, ensuring sellers have the infrastructure and visibility needed to scale profitably.

In this conversation, we discussed about the state of AI in logistics, especially the shipping industry and what lies ahead:

• Where AI is creating measurable value across logistics and shipping operations
• The balance between automation, human expertise, and operational reliability
• What the next generation of AI-powered logistics platforms will look like

In Conversation with Industry Insiders

Q1. From your perspective, how has the logistics and shipping industry evolved over the past decade, especially with e-commerce growth?

I’ve been in shipping since 2001, so I’ve watched the whole thing happen. The biggest change isn’t volume; it’s who’s doing the shipping. When I started, shipping was a back-office function for a small number of big companies. Today, the growth is coming from millions of small operators, independent merchants, direct-sales consultants, marketplace sellers, and people running brands out of their garage.

That broke two things at once. The carriers were built for B2B pallets. Now they’re hauling individual parcels at consumer scale, and the rate structures show it: dim weight, surcharges, peak fees, residential adjustments. Most of that didn’t exist in the form we deal with now twenty years ago. The other shift is on the customer side. A shipping label used to be operational. Now it’s the last brand moment before the customer opens the box. If it’s slow or messed up, your brand takes the hit, not the carrier’s.

The frustrating part for most of my career has been that the complexity got way ahead of the tools. Operators were running shipping on instinct and spreadsheets because anything better was too expensive to maintain. The use of artificial intelligence in logistics is the first thing I’ve seen that actually closes that gap. And what’s interesting is it’s reaching the small operators first; that’s never happened with any previous wave of technology in this industry.

Q2. Where do you currently see AI making the most practical impact in logistics operations today?

The real AI work is happening in the boring parts of the job, not the headlines. Five places I’d point to.

• Invoice reconciliation. AI reads the carrier invoice line by line, ties each charge back to the shipment it came from, and flags when what you got billed doesn’t match what you were quoted. That used to be a finance project once a quarter. Now operations sees it every day.
• Picking the right box. Logistics applications built using artificial intelligence and computer vision can tell you which box to use for which product to keep your dim weight down. That decision used to sit in the warehouse with no connection to your cost model. Now it’s connected.
• Rate and service selection. Multi-carrier rate shopping has been around for years. What’s new is AI picking your service level based on what will actually deliver on time, what surcharges will hit, and what your customer expects, not just the published rate.
• Audit after the fact. Re-weighs, address corrections, and late dim adjustments; that used to disappear into the finance close. AI finds it, sorts it, and routes it for dispute on its own.
• Customer service. About two-thirds of “where’s my package” tickets are the same question. AI handles those cleanly, so your team can spend their time on the ones that actually need a human.

If you back up, the common thread is this: AI is finally letting an operator see what their shipping is really costing them, in time to do something about it. For a small or mid-sized shipper, that was theoretical five years ago. It isn’t anymore.

Q3. There is a lot of momentum around AI in logistics right now, and many platforms are adopting it for optimization. But in your experience, what is actually working versus what is still hype?

I’ll tell you what I see. Invoice anomaly detection, claims automation, dim-weight estimation, support deflection on tracking, and routing inside clear rules are some of the use cases of AI which are the most functional. All of it is saving real money for the operators using it now.

Here’s what is hype though: “agentic supply chains,” “self-optimizing networks,” “AI will replace your warehouse planner.” The marketing is years ahead of what the technology can actually do. AI can draft a good decision. It can’t own a decision when your carrier changes pricing rules on a Tuesday, weather hits on Wednesday, and a customer wants a custom SLA on Thursday.

Simplest way I’d put it: AI as a co-pilot for someone who knows what they’re doing, works great. AI as the pilot, making decisions with no one watching, breaks the first time something changes. In shipping, things change every week.

The operators getting this right aren’t picking the most ambitious tool on the market. They’re picking the work where the same answer is right 95% of the time, automating that, and sending the other 5% to a person. That’s the whole formula. The autonomous-warehouse pitch sells conference tickets and not much else.

Q4. Zooming into day-to-day operations, what are the biggest operational bottlenecks in logistics workflows today that AI can realistically solve?

Four big ones, in my experience.

• Rate shopping. Picking the right service for each package without somebody clicking through carrier portals all day. AI handles the simple version, and now it handles the dynamic version too, adjusting to surcharges and SLAs in real time.
• Exceptions. Address fixes, failed deliveries, RMAs, and claims paperwork. This stuff eats way more operator time than its share of the volume. AI handles the routine work. The judgment calls still need a person.
• Customer service triage. Separating “I’m worried, where is it?” from “something’s actually wrong”; AI does that well, and it frees up your people for the cases that matter.
• Cost visibility. This one’s underrated. The gap between what your shipment was quoted and what you actually got billed has been an unsolved problem at scale for as long as I can remember. AI closes it. That changes shipping from something you review every quarter to something you manage every day.

What AI can’t do today: anything physical. Pick paths, packout speed, dock scheduling at peak; those are still labor and process problems, and they will be for a while.

Q5. At the same time, logistics is still a very human-heavy industry in many ways. How do you balance automation with the need for human support in shipping and fulfillment operations?

Two things are true at the same time. AI is good at scale and consistency. People are good at judgment when things shift. The question is which kind of work you’re looking at.

A rule that’s worked for us: if the same answer is right 95% of the time, automate it. The other 5% goes to a person. The mistake I see most often is replacing a human with a chatbot that can’t escalate. Customer hits an edge case, the bot loops, the customer’s gone, and whatever you saved on headcount is gone with them.

Shipping is high-stakes for the customer. A missed delivery is a brand moment, sometimes a public one. So the bar for keeping a person in the loop is higher here than in other industries. The setup that ages well: AI drafts, a person approves. AI suggests the routing, and a person can override. That also keeps your operators sharp, which matters for the next decision the AI hasn’t seen yet.

Q6. Behind all of this, data plays a critical role. What role does data quality and system integration play in enabling AI-driven logistics platforms?

AI is a multiplier on data quality. Bad data, faster bad decisions. That’s it.

Three things go wrong on repeat. Stale carrier rate cards, a model trained on last quarter’s rates, is losing money by Wednesday. Inconsistent SKU dimensions, your dim-weight math is only as good as the box size and weight you’ve recorded. And order-system fragmentation, the same customer’s address shows up three slightly different ways across three platforms.

The real moat isn’t the model, the real MVP is the data engineering and plumbing behind. The companies winning right now aren’t the ones with the fanciest AI. They’re the ones whose systems let the model see orders, inventory, carriers, customers, and exceptions in one place. Cleaning up a data pipeline beats buying a better model nine times out of ten.

There’s a bigger point underneath this that most of the coverage misses. AI’s biggest impact on shipping isn’t faster decisions for the Fortune 500. It’s that the data and integration work that used to need a team of engineers is now affordable for a small operator. The ceiling AI is raising is the one over the small shipper, not the big one. That’s the change I’d watch.

Q7. Where do logistics companies typically struggle the most when trying to integrate AI into existing shipping infrastructure?

Three patterns I see over and over.

• Legacy systems. A lot of WMS and TMS platforms were built before APIs were standard. Strapping AI on top of duct-taped integrations means the model is reading the wrong data, late.
• Carrier variability. Carriers themselves are all over the map on API quality and rate transparency. Your AI is only as good as the carrier responses it can rely on, and most AI vendors don’t talk about that ceiling.
• Org structure. AI initiatives end up in IT. But the real wins are in operations, customer service, and finance: the teams whose work actually changes. If they’re not in the room when you design the rollout, the rollout stalls. The fix is unglamorous: cross-functional ownership from day one, and real change management for the people using the tool.

Q8. Looking a bit further ahead, there is growing conversation around autonomy in supply chains. How close are we to fully autonomous logistics workflows in e-commerce fulfillment?

Further out than the conference circuit will tell you. Single steps, such as printing a label, picking a rate, sending a tracking update, those are solved. Multi-step coordination across carriers, customs, weather, and customer preferences is not.

My read: in two to three years, AI as the default operator with people on the exceptions is realistic. That’s where most of the industry will land. Real end-to-end autonomous orchestration is five years out at least, and the blockers aren’t AI capability. They’re physical-world variability and trust. Operators, carriers, and customers all have to agree to hand over decisions that cost real money when they go wrong.

eCommerce fulfillment has a long-tail problem on top of all that. The data maturity you need for autonomy isn’t where most small and mid-sized shippers are. They’ll get value from AI as a co-pilot for years before they’re ready to give it the keys. And the co-pilot version is what’s actually moving margins right now anyway. That’s the right place to focus.

Q9. Of course, with scale comes risk. So, what risks should logistics companies be cautious about when deploying AI at scale?

Four things I’d flag.

• Confident-wrong errors. Your AI routes ten thousand packages to the wrong carrier before anyone notices, because the system thinks it’s working.
• Drift. The model was trained on last quarter’s rates and surcharges. Carriers update those weekly. Without ongoing retraining or guardrails, your optimization quietly degrades and you don’t see it until the finance review.
• Customer trust. An automated tracking message that’s wrong damages your brand faster than a slow human response would.
• Cost-only optimization. If you tune the AI to grab the cheapest label every time, you’ll end up shipping ground when the customer expects express. The cost line looks great. The retention line doesn’t.

The fixes aren’t glamorous. Run in shadow mode before you turn it on for real. Audit a sample of decisions. Set anomaly thresholds. Have a kill switch. Make sure there’s a clear path to a human. Treat the deployment like a launch, not a one-time install.

Q10. To wrap this up, if we look 3–5 years ahead, what does the future of AI-powered logistics platforms look like to you?

Two predictions I’d hold together.

First, the platform layer wins. If you’re still running shipping decisions across four disconnected systems in 2030, you’re paying a margin penalty your competitors aren’t. AI compounds on integrated data and stalls on fragmented data. That’s a structural advantage you either build or lose.

Second, and this is the one I care most about, AI levels the field for small operators. For most of my career, good shipping intelligence has belonged to enterprise shippers with the volume to justify a full team. Rate optimization, dynamic routing, real audit data, predictive analytics, all enterprise tools. In the next three to five years, the solo merchant or direct-sales consultant working from her kitchen table is going to have the same shipping intelligence and software backing logistics and transportation ops that used to take a full logistics department.

The question I’d put to anyone in this industry: it’s not whether AI will run logistics. It’s whether the small operator gets access to the same AI as the enterprise. The answer to that decides whether eCommerce stays open to new entrants or quietly consolidates around the players big enough to build their own AI. I’m betting it stays open, but only if the platforms serving small operators take AI seriously, right now.

Radixweb Perspective on AI in Logistics

One of the strongest takeaways from Kyle’s insights is that the future of logistics is not about automating isolated tasks. It is about building intelligent operations where data, workflows, and decision-making work together seamlessly. This aligns closely with what we see across logistics and supply chain projects. Organizations often focus on AI capabilities, but the real impact comes from connecting fragmented systems and then integrating AI to improve data visibility and embedding intelligence into everyday operations.At Radixweb, we help businesses move beyond experimentation by building AI logistics solutions that integrate with existing platforms, automate complex workflows, and support faster, more informed decisions. Our experience shows that the greatest returns come when AI becomes part of the operational backbone rather than a standalone tool. If you're evaluating opportunities to modernize shipping, fulfillment, or logistics operations, schedule a consultation with our experts to discuss your intelligent logistics roadmap.