I’ve spent the better part of three decades in supply chain technology
I’ve watched this industry go from green-screen terminals and floppy disks to cloud-native platforms and generative AI. I’ve lived through client-server, survived the Y2K hysteria (twice, if you count the consultants who reinvented it for 2000), watched mobile change everything, cheered when SaaS finally freed us from on-premise upgrades that took longer than actual construction projects, and now I’m watching AI rewrite the rules again.
It actually started with a conversation
I was riffing with Juan Cora of Aquatio Software — we were swapping notes about our kids and where they’d inherited their music taste. Somewhere in that conversation, we realized: our children had moved on to whatever the kids are listening to now, but we were still classic rock guys. And then it hit me that the same thing was true of our supply chain careers. The warehouse management system (WMS) industry has been playing the same hits for a long time.
You know the one. It plays the same 40 songs on rotation. “Free Bird.” “Stairway to Heaven.” “Don’t Stop Believin.'” Every. Single. Day. It’s comfortable. Familiar. A little nostalgic. And absolutely, completely, hopelessly stuck in 1987.
The warehousing software world has its own classic rock station — and it’s been playing the same hits for a long time.
Let me take you on a tour
My career has been a front-row seat to every major technology shift in supply chain. And I mean that literally — in my entire career, every company I’ve worked for has been a supply chain company, or had supply chain companies as its primary clients. I’ve never left the building. Supply chain has been my whole career — and warehousing, specifically, has been the thread that runs through most of it. I’ve worked across the full spectrum — PR and analyst relations, supply chain planning, B2B marketplace technology, enterprise WMS, ERP roll-ups, cloud-native startups, supply chain risk intelligence, and now commerce intelligence at Deposco. I’ve been a CMO, a VP of Marketing, a storyteller, and occasionally the person who inherits a PowerPoint deck that uses Comic Sans unironically. (We fixed it.)
And now Deposco — where we’re building what I genuinely believe is the supply chain platform for the AI era. More on that in a moment.
The technology eras: a brief, biased history
If you want to understand why so many WMS platforms are stuck, you have to understand the eras they were built in — and how each one created legacy debt that compounds over time.
Era 1: Client-Server (the mullet era)
Client-server architecture was the mullet of software: business in the front, party in the back. You had a powerful server doing the real work, and a “thin” client on the desktop pretending to be useful. Most of the WMS platforms built in the 1990s and early 2000s were architected this way. They were powerful for their time. They could handle warehouse operations at scale. But they required dedicated servers, dedicated IT staff, dedicated upgrade weekends that everyone dreaded, and a tolerance for pain that bordered on the spiritual. The era’s anthem: “Money for Nothing” by Dire Straits. Powerful machinery, impressive output — and you were absolutely going to pay for it.
The dirty secret about client-server WMS? Many of them are still running today. In 2026. With the same core architecture. The same database schemas. The same logic. Lovingly maintained by people who are either deeply committed or quietly updating their LinkedIn profiles.
Era 2: Mobile-Enabled (we added an app!)
The mobile era arrived and WMS vendors scrambled. Most of them didn’t rebuild for mobile. They wrapped. They bolted. They created “mobile-enabled” versions of their desktop UIs that technically ran on a tablet the way a Cadillac technically fits in a compact parking space. Possible? Sure. Graceful? Absolutely not. Jackson Browne had it right: “Running on Empty.” Still moving, technically. But anyone paying attention could see the tank wasn’t going to last.
Mobile was a forcing function that revealed which vendors had modern, API-first architectures and which ones had a spaghetti of legacy code held together by custom integrations and prayers. Spoiler: there was a lot of spaghetti.
Era 3: The Cloud (“cloud-ready” vs. cloud-native)
The cloud era created one of the greatest marketing sleights of hand in enterprise software history: the distinction between “cloud-ready” and “cloud-native” was systematically ignored by anyone who had something to sell. Radiohead called this era perfectly with “Fake Plastic Trees” — looks entirely like the real thing, right up until it doesn’t.
I’ve seen firsthand what a truly cloud-native WMS looked like. It was architected for elasticity, multi-tenancy, and continuous deployment from day one. Contrast that with legacy vendors who took their on-premise systems, put them on AWS or Azure, and declared victory. “We’re in the cloud now!” Sure. But so is a filing cabinet if you put it in a data center. That doesn’t make it a cloud application.
The cloud-native vs. cloud-washed distinction is not academic. It determines how fast you can scale, how easily you can integrate, how often you get new features, and — critically — how ready you are for what comes next.
Era 4: SaaS (finally, someone else’s problem)
SaaS was the gift that kept giving — for customers, at least. No more upgrade cycles that consumed six months and required a dedicated project manager, three consultants, and a catering budget for the war room. Instead: continuous updates, shared infrastructure, and a pricing model based on value delivered rather than licenses purchased and forgotten. The Beatles nailed the feeling: “Here Comes the Sun.” After years of on-premise winters, it finally felt like relief.
But here’s the thing about SaaS: calling yourself SaaS doesn’t make you SaaS any more than putting “artisan” on a menu makes the bread hand-crafted. True SaaS means the product evolves continuously for all customers simultaneously. A lot of WMS vendors offer “hosted” or “managed” versions of their on-premise products and call it SaaS. It’s not. It’s a costume.
Era 5: AI — and this is where the classic rock station gets existential
And now we’re here. Generative AI. Agentic AI. Multi-agent systems that can reason, plan, execute, and learn. AI that doesn’t just report what happened but predicts what will happen and recommends what to do about it. Bob Dylan wrote the soundtrack fifty years early: “The Times They Are A-Changin’.” He was talking about a cultural revolution. We’re talking about a platform revolution. The stakes for your warehouse are remarkably similar.
This isn’t an incremental upgrade. This is a platform shift. And platform shifts expose every shortcut, every workaround, every piece of legacy debt that was papered over in previous eras.
The classic rock station is getting a request to play something from 2026. And it literally does not have the music.
The classic rock WMS problem: greatest hits only
Let me be specific about what “stuck on the classic rock station” actually means in WMS terms, because this isn’t just a fun metaphor — it’s a real operational and strategic risk.
Classic rock WMS platforms were built to manage what happened — receiving, putaway, picking, packing, shipping. Transactions. Events. Historical records. They are excellent at telling you what your warehouse did. What they are not designed to do is tell you what your warehouse should do next, at scale, in real time, using intelligence derived from multiple data sources simultaneously.
Generative AI and agentic AI require: clean, structured, accessible data at the event level; modern API-first architecture that allows AI agents to read and write in real time; cloud-native infrastructure that can scale compute dynamically when AI workloads spike; and continuous deployment pipelines so AI models can be updated as they learn.
Go look at the architecture of a 1999-vintage WMS that has been “cloud-enabled.” It has a monolithic database optimized for batch processing. It has APIs that were an afterthought, not a design principle. It has an upgrade model where new versions are discrete events, not continuous flows. And it has a data model that was never designed to feed a machine learning pipeline.
You can’t bolt generative AI onto that any more than you can add a turbocharger to a horse-drawn carriage. The horse doesn’t appreciate it, and the carriage wasn’t designed for the stress.
The oldies channel problem: when support gets sunset
Here’s where it stops being funny and starts being genuinely urgent.
Several legacy WMS platforms are reaching — or have reached — end-of-life status. Vendors are sunsetting products, dropping support tiers, and quietly steering customers toward “new” platforms that are sometimes just the old platform with a new logo and a price increase. In some cases, the migration path the vendor is offering isn’t a migration path at all — it’s a reimplementation. Start from scratch. Re-train your team. Re-integrate your ERP. Re-configure your workflows. Goodbye, sunk costs. Hello, new sunk costs.
If you’re running a WMS that is approaching end-of-support, you are not just sitting on legacy technology. You are sitting on a liability. Every month you stay on an unsupported platform is a month of security risk, integration debt, and growing incompatibility with the ecosystem around you — your ERP, your carriers, your automation vendors, your e-commerce platform, and yes, your AI tools.
The classic rock station isn’t just stuck. In some cases, it’s being taken off the air. And the customers who depended on it are left holding a radio with no signal.
What “modern” actually means (and why it matters now)
I’ve worked for enough supply chain software companies to be deeply skeptical of the word “modern.” Every vendor calls their platform modern. It’s the supply chain software equivalent of a restaurant calling its food “fresh.” Technically true. Functionally meaningless.
So let me tell you what modern actually means in 2026, based on three decades of watching what separates the platforms that age well from the ones that don’t.
Modern means cloud-native — not cloud-washed
Built for multi-tenancy, elasticity, and continuous deployment from the ground up. Modern means API-first. Not APIs as an afterthought or an add-on module. APIs as the primary interface through which everything talks to everything else. Modern means a data architecture designed for intelligence — structured, event-level, real-time, and accessible to AI and analytics tools without heroic ETL work. And modern means an AI strategy that is embedded in the product roadmap, not stapled to the marketing deck.
How do you cut through the AI noise? Guide to supply chain AI terminology, questions to ask, and where to start.
How we built our Supply Chain Intelligence platform at Deposco
We call it SCI, powered by our Felix multi-agent AI — built on exactly these principles: cloud-native, not cloud-washed.
Felix isn’t a chatbot with warehouse knowledge. It’s a multi-agent AI system that can reason across your supply chain data, identify optimization opportunities, and take action — not just make recommendations. The difference between “AI that shows you a dashboard” and “AI that actually works in production” is the difference between a weather forecast and a self-driving car.
That’s not possible on a classic rock WMS. It requires the architecture underneath. And that architecture takes years to build — which is exactly why starting the migration conversation now matters more than most people realize.
So what should you do?
If you’re a supply chain leader reading this while quietly wondering whether your WMS is on the classic rock station — or worse, the oldies channel heading toward format change — here’s my honest, 30-year-career-informed advice:
First, ask your WMS vendor a very direct question
“What is your AI roadmap, and what specifically in your architecture supports it?” If the answer involves the phrase “we’re evaluating” or “we have a partnership with [insert AI company here],” be worried. That’s a vendor who is planning to bolt AI onto a platform that wasn’t designed for it. The partnership announcement is the press release they’ll put out when they run out of time.
Second, understand your support horizon
If your platform has a known end-of-life date, you need to be planning your migration now — not when the sunset hits. WMS migrations are not fast. They are not cheap. And they are much harder to execute under duress than under a planned, rational timeline.
Third, when you evaluate alternatives, look beneath the demo
Every WMS looks great in a demo. Ask about the deployment model, the upgrade frequency, the data architecture, the API documentation, and — critically — reference customers who have implemented AI use cases in production. Not pilots. Not proofs-of-concept. Production.
The supply chains pulling ahead in 2026 aren’t just faster. They’re smarter. And smart requires a platform that was built to be smart — not one that’s trying to learn the new songs while still managing requests for “Free Bird.”
About the Author
Todd Craig is the Chief Marketing Officer at Deposco. He has spent 30+ years in supply chain technology marketing, with stops across supply chain planning, WMS, ERP, risk intelligence, and cloud-native software — the full tour. He holds a degree from the University of Alabama — Roll Tide — and still has strong opinions about both WMS architecture and the classic rock canon.