Retail's dirty little secret? Most of our systems don’t talk to each other.
For years, we’ve bolted on new tech in hopes it would create clarity. Instead, we ended up with disconnected software—each tool with its own data, interface, and logic. The warehouse couldn’t see labor needs. POS had no idea what was stuck in the receiving area. Scheduling was a guessing game.
Want to know retail's dirty little secret? Most of our systems don’t talk to each other.
For years, we’ve bolted on new tech in hopes that it would create clarity. Instead, we ended up with disconnected software—each tool with its own data, interface, and logic. The warehouse couldn’t see labor needs. POS had no idea what was stuck in receiving. Scheduling was a guessing game.
At Manhattan Associates’ annual conference, Momentum 2025, Sanjeev Siotia, CTO of Manhattan Associates, put it simply, “You’ve got amazing software. But they’ve each been built like highly intelligent people sitting in separate rooms with the door closed.”
That’s the real pain point AI must solve—not just automation, but integration.
Retail Complexity Outpaced Retail Tools
Retail has evolved from simply stocking shelves to orchestrating vast ecosystems—from vendors to delivery docks to customer doorsteps. And today’s consumer expects real-time answers and frictionless service at every step.
The problem? Legacy systems weren't designed to handle this level of real-time complexity. And worse, each department runs its own tech stack, unable to collaborate beyond CSV exports or Slack messages.
“Human-computer interactions have come a long way,” Sanjeev said, “but they’ve remained largely transactional. What we need now are systems that can reason with us, act with us, and grow with us.”
Fragmented AI Is Just Another Silo
In a fragmented world, solving one customer issue would require four disconnected tools:
- One to alert the customer,
- One to create a new shipping label,
- Another to update inventory,
- And yet another to trigger follow-up actions.
Sanjeev explained that Manhattan sees it differently: rather than separate tasks handled by different systems or people, they treat it as a single, intelligent action. “Instead of one system alerting the customer, another creating a shipping label, and another updating inventory,” he said, “an agent can now handle it all automatically, reliably, while always looking for the next best way to help.”
That shift—from “many tools doing a single job” to “one agent designed to help”—is the breakthrough. It’s not automation. It’s coordination.
“AI isn’t helpful if it lives in a bubble,” Sanjeev warned. “An agent needs to think, see, and do—all in the context of your entire operation.”
He explained this as the three required proficiencies of AI agents:
- Think: Understand multi-step goals and how to achieve them.
- See: Have real-time access to all your operational data.
- Do: Execute the right actions, without human handoffs.
This holistic capability isn’t available from bolt-ons. It’s only possible when your AI lives inside your platform, not bolted onto it.
Unified AI Agents Built into Manhattan Active
Manhattan didn’t just add AI to their platform—they rearchitected their platform to become truly agentic.
These agents don’t just suggest—they act. They don’t wait for a human to interpret a dashboard—they execute in real time. And more importantly, they work seamlessly together.
Sanjeev showed a live demo featuring Agent Joe, a labor optimization agent that could dynamically reassign employees based on shifting work volumes:
He asked the computer, “Can you tell me if any departments are running behind?”
The computer responded, “Yes, packing is one hour behind. You can reassign three pickers finishing early.”
It then took stock of who was on the schedule, and then automatically sent them a note when they were finished to go help with packing.
That wasn’t science fiction. That’s production-ready software that many Manhattan clients are already using. Natural language. No prompt necessary, just the goal of what you want to know.
Agent Foundry: Retailers Get to Build Their Own Intelligence
Manhattan’s customers will get access to Agent Foundry, the same environment Manhattan uses to build its AI agents.
“If you know your business, you can build an agent,” he said. “And if you don’t know how? We can build agents for you.”
That democratization of AI is critical. It means innovation isn’t bottlenecked by developers or Manhattan’s roadmap. If your floor team wants a packing assistant who knows their workflow, they can build it.
One load planning team used this to cut a 3-hour daily task down to minutes by defining simple rules in natural language.
“That’s the power of platform-native intelligence,” Sanjeev said. “Agents inherit the entire ecosystem they’re built in. They’re born integrated.”
Consulting Agents: Removing Complexity
Sanjeev also revealed consulting agents that can read a business requirements document and configure the system accordingly.
“We want our systems to understand your business like a consultant,” he said. “Eventually, you’ll just describe what you want—and it’ll build itself.”
That’s not a fantasy. It’s a roadmap that’s already in progress.
Customer-Facing Displays: Transparency That Builds Trust
Unified AI doesn’t just help your teams behind the scenes- it enhances the shopper’s experience at the point of sale, too.
Manhattan showcased a new customer-facing display that syncs directly with the POS. As shoppers check out, they can see their:
- Loyalty points accumulate
- Available coupons to apply
- Basket being rung up in real time
That visibility creates confidence. No guesswork. No surprises. No more squinting at receipts or guessing what promotion applied. It’s trust, built in real time. When systems are unified, the customer’s experience becomes unified too. Just a clear, trust-building interaction at a moment that too often feels rushed or opaque.
When systems are unified from warehouse to front-end checkout, you create harmony across the whole experience. And it’s small details like this that reinforce a big transformation.
Why This Is a Turning Point for Retailers
I’ve been in retail for decades. I've seen every so-called breakthrough: barcode scanners, ERP rollouts, cloud POS. But most of them still relied on humans pulling the right levers.
This feels different. Unified AI agents don’t just reduce effort—they reduce friction. Between departments. Between platforms. Between decisions.
Here’s the big idea:
You don’t need more dashboards.
You need intelligence that acts in real time.
You need agents that learn and operate as one.
And Manhattan is building the tools that finally make that possible.
The Next Great Supply Chain Revolution Isn’t Faster. It’s Frictionless
Sanjeev put it best: “We’re not here to replace your people. We’re here to elevate them—to give them tools that can think and act so they can focus on what matters most.”
If you're still using systems that can't talk to each other, can't learn, and can't act without intervention, you’re operating at a disadvantage.
Unified AI agents are here. They work together. And they’re already being used to reduce cost, save time, and unlock agility at scale.
And the benefits aren’t limited to back-end teams. Shoppers feel it too.
At Manhattan’s Discovery Center, a customer-facing display showed shoppers their loyalty points, usable coupons, and their basket being rung up—live, in real time.
Your Next Steps
Ask: What repetitive decisions are eating up our team’s time?
Evaluate: Are your systems built to think, see, and do, or just display?
The days of “see no data, hear no signals, speak no solutions” are over.
It’s time for your systems to start working together.
And with AI, they finally can.
