Retail 2026: When AI becomes the operating system – News

As the National Retail Federation’s annual conference takes over New York this week, the conversations we’re having across News, theCUBE, theCUBE Research and NYSE Wired all point to the same conclusion: Retail is entering a once-in-a-generation transformation. The signals are everywhere — from store floors to supply chains to boardrooms — and they tell a single story.

Retail has crossed a threshold.

What began as experimentation — chatbots, recommendation engines, demand forecasts — has become something far more consequential. By 2027, artificial intelligence will no longer be a layer on top of retail operations. It is the operating system. It runs merchandising. It orchestrates supply chains. It powers marketing. It assists employees. It increasingly makes decisions in real time.

This isn’t a metaphor. Databricks Chief Scientist Jonathan Frankle frames it plainly: “AI is just a different kind of computing.” In other words, this is not a feature wave — it’s a platform shift.

The defining change isn’t whether retailers are investing in AI. That question is settled. The real story is how AI is being deployed at scale— and what separates companies that merely adopt from those that build durable advantage.

As CIO advisor Tim Crawford puts it, “We’re moving past chatbots into systems that actually take action.” That move — from suggestion to execution — forces architectural and governance changes that look far more like an operating system upgrade than a software rollout.

This is the year retail moves from tools to systems.

The rise of the composable AI stack

Retailers have learned a hard lesson: their businesses are too unique for one-size-fits-all intelligence.

Every retailer carries its own DNA — assortments, pricing strategies, demand curves, regional behaviors, loyalty dynamics. Generic models can’t fully capture that complexity. The result is a quiet but decisive move toward open, composable AI stacks.

Stephen Orban of Databricks observes that NRF has become “as much a tech show as it is a retail show.” The implication is clear: retailers are no longer shopping for monoliths. They are assembling ecosystems—choosing from marketplaces of models, tools, and partners that can be mixed, matched, and tuned to their business.

Open models and frameworks now sit at the core of modern retail architectures. They allow teams to fine-tune intelligence on proprietary data, adapt quickly to new use cases, and avoid long-term dependency on a single vendor’s roadmap. What began as a developer preference has become an executive strategy.

AI is no longer something you “buy.” It’s something you shape or steer.

But composability without discipline becomes fragility. Crawford offers a sharp warning: “You can’t build a Frankenstein of data architecture.” The winners are pairing openness with rigor — treating models like infrastructure: customized, governed, and deeply integrated into how the business runs.

From insight to action: The agentic turn

The most profound change underway is the move from analytical AI to agentic AI. For years, systems told humans what was happening. Now, they are starting to act.

Salesforce’s GM of Retail Cloud Nitin Mangtani describes an “agentic layer” that goes “beyond data retrieval… all the way to the action.” In practice, this means systems that don’t just flag an inventory imbalance — they correct it. They don’t just surface a churn risk — they trigger a tailored response. They reason across data, plan sequences of steps, and execute tasks across multiple systems.

Mike Micucci, CEO of Fabric, sees this shift on the ground. He describes “agentic AI systems” built around role-specific agents for operators and merchandisers. These aren’t dashboards. They are digital coworkers — automating tasks, coordinating across systems, and making decisions.

This is where advantage shifts from “better dashboards” to autonomous execution. Retail is becoming a machine-speed business. Human teams can no longer manually orchestrate thousands of micro-decisions across channels and touchpoints. Agents can.

The competitive frontier is no longer who has the most data—but who can turn intent into action fastest.

Fixing the inside before transforming the outside

Interestingly, the first wave of agentic transformation is happening inside the enterprise.

Micucci calls it directly: “The real opening up is the operational aspect.” Inventory allocation, fulfillment optimization, exception handling — these are where agents deliver immediate, compounding returns.

Retailers are using artificial intelligence to unlock what has long been trapped: institutional knowledge buried in documents, legacy systems and individual employees. Agents now retrieve policies, summarize contracts, guide workflows and assist teams in real time.

Operations, merchandising, finance, HR, store management — every function is being rewired with intelligent assistants that reduce friction and cognitive load.

Before AI reshapes the customer experience, it is fixing the internal plumbing. This is not glamorous work. But it is foundational. The retailers who get this right build velocity. They move faster, onboard quicker, and scale expertise across the organization.

AI becomes the nervous system of the enterprise: the operating system.

That said, the path is not strictly linear. Thea Myhrvold shows how operations and experience can advance together, using human-in-the-loop and multimodal AI to elevate customer engagement even as back-office systems modernize. Many retailers will run both tracks in parallel.

The small get strong

One of the most surprising dynamics of this era is how aggressively smaller companies are leaning into agentic systems.

Myhrvold describes getting clients live “in less than a week,” using AI to empower store associates and deliver high-touch, human-led experiences without enterprise-scale engineering. What once required massive information technology teams now fits inside lean organizations.

Although large enterprises will still dominate in analytics and foundational platforms, smaller retailers are experiencing something different: a chance to punch above their weight.

With fewer layers and tighter teams, they deploy autonomous systems as force multipliers. Agents handle tasks that once required entire departments. Execution scales without proportional headcount. What used to demand “enterprise scale” now fits inside a lean organization.

AI is quietly flattening the industry.

In a world where autonomy replaces bureaucracy, speed becomes a weapon. And small, focused players suddenly look dangerous.

Commerce becomes conversational — and continuous

Digital commerce remains the gravitational center of AI investment, but its nature is changing.

Personalization is no longer a feature. It is the baseline. Product catalogs enrich themselves. Shopping journeys adapt in real time. Assistants guide discovery, resolve uncertainty and anticipate intent. The storefront becomes dynamic — reshaped continuously by signals from behavior, context and history.

Myhrvold captures the shift succinctly: “We’re moving from scroll-based to goal-based shopping.” Computer vision and multimodal models let customers express intent naturally — showing, speaking, asking — and receive outcomes, not just results.

Mangtani ties this directly to execution. Associates can transact anywhere. “Endless aisle” becomes real as cloud inventory is exposed across channels.

This is the emergence of agentic commerce. AI doesn’t just recommend. It negotiates attention. It curates. It learns. It responds. It becomes part of the shopping experience itself.

The retailers that win here don’t bolt AI onto the funnel. They reimagine the funnel as a living system.

Stores go deep, not wide

Physical retail is not being automated wholesale. Instead, it is becoming more precise.

Where AI is deployed in stores, it is focused on insight: understanding traffic patterns, dwell time, conversion behavior and cross-channel journeys. Now, digital and physical experiences will converge into a unified customer view. There will be a hyperconvergence at the edge, and retail will feel it immediately — especially as telcos and carriers change their role in the value chain.

The goal isn’t to replace retail workers, but to give them superpowers. Stores become sensors. Floors become data. The physical world becomes legible. Physical AI will be a key driver here.

Brick-and-mortar doesn’t disappear. It gets smarter.

Supply chains become strategic weapons

Few areas reveal AI’s impact more clearly than supply chain. Volatility is now permanent — weather, geopolitics, labor, transportation, demand shocks. What used to be optimization problems are now survival problems.

Micucci describes AI “inter-meshing inventory, orders and products,” with agents assisting operators to continuously rebalance allocation and fulfillment. Intelligence becomes margin. Precision becomes conversion.

AI is turning supply chains from cost centers into competitive weapons. Systems forecast, simulate, reroute and rebalance in near real time. They don’t just reduce cost — they protect revenue, preserve trust, and absorb disruption.

Retailers are discovering that resilience is a customer experience. When shelves are stocked and deliveries arrive on time during chaos, loyalty compounds. This is why investment here is accelerating. The ROI is no longer theoretical. It’s visible in margins and market share.

The physical world awakens

Robotics and physical AI remain early — but their trajectory is unmistakable.

Simulation-driven logistics, autonomous movement, robotic handling — these are no longer science projects. They are being stitched into the fabric of distribution centers and warehouses.

As digital intelligence meets physical execution, retail builds adaptive systems that respond to the real world, not just models of it.

The real constraint: people

Data is no longer the primary bottleneck. Infrastructure is no longer the primary bottleneck.

People are.

The industry is pouring capital into AI. But the ability to design, deploy, govern, and operate intelligent systems is scarce. Execution — not ambition — is the limiter. Strategy isn’t the risk factor. Execution is.

The retailers that invest in talent, training, and organizational design will compound their advantage. The rest will buy tools they cannot fully use.

AI is not plug-and-play. It is a new operating model.

Inference is the new battleground

As AI moves into production, economics comes into focus.

Every recommendation, forecast, and decision carries a cost. At retail scale, inefficiencies multiply fast. Latency becomes experience. Throughput becomes revenue. Architecture becomes strategy.

Mangtani frames it in business terms: “The moment you reduce friction on commerce, the pie increases.” That is why inference performance, cost, and reliability are moving into boardroom conversations. These are no longer engineering tradeoffs — they are growth levers.

This is driving a new class of hybrid stacks — mixing proprietary systems with open components — to optimize performance at scale. Inference is no longer a technical footnote. It is a board-level conversation.

This is where advantage becomes durable.

From adoption to advantage

Retail is entering its decisive phase.

The foundations are in place. The experiments are over. The systems are forming.

The shift is unmistakable:

• From experimentation to execution
• From tools to autonomous systems
• From back-office efficiency to front-line transformation
• From isolated models to end-to-end intelligence

The retailers that close the talent gap, operationalize agentic systems and master the economics of inference will not simply improve efficiency.

They will redefine what retail is.  The next 12 months won’t determine who is “doing AI.”  They will determine who is leading with it.

Photo: NRF