A shopper walks into a store with a specific need. Maybe they’re fixing an irrigation system, planning a meal, or trying to resolve a membership issue. Instead of searching aisles or waiting for help, they walk up to an assistant and start a conversation. The assistant understands the store, the inventory, and the context of the question. It responds immediately, in the shopper’s preferred language, and guides them to what they need next. But here’s the catch; the assistant is virtual.
That experience is no longer theoretical. It’s a glimpse of where retail AI is headed and why the store itself has become the most important place for intelligence to run.
The reason is simple: where data is processed is changing dramatically. According to Gartner, by 2027, an estimated 75% of data will be processed outside of traditional data centers. For retail, that shift isn’t abstract. It reflects a growing need for intelligence to live closer to customers, associates, and real-world interactions.
A Glimpse of Retail AI Where It Actually Happens
What makes this kind of interaction possible isn’t just better AI models. It’s where those models run.
Retail use cases like conversational assistance, personalization, video analytics, and inventory intelligence all depend on real-time decision-making. Latency is one part of the equation, but it’s not the only challenge retailers face. Reliability matters. When AI relies on constant round trips to a centralized cloud, even small delays can disrupt the experience. Bandwidth constraints, connectivity interruptions, and rising data movement costs can quickly turn promising use cases into operational headaches.
There’s also the question of data sovereignty. Much of the data generated inside the store (video feeds, customer interactions, operational signals) is sensitive by nature. Retailers increasingly want control over where the data is processed and how it’s handled, rather than pushing everything to a distant cloud or enterprise data center.
That’s why more retailers are rethinking the role of the store. It’s no longer just a source of data. It’s becoming an execution environment for AI — where decisions happen locally, instantly, and in context while training and optimization occur centrally. This approach improves responsiveness, strengthens resilience when connectivity is constrained, and gives retailers greater control over their data.
This shift allows AI to support everyday retail moments: answering questions accurately, helping newer employees fill knowledge gaps, and removing friction from interactions that used to rely on static kiosks or hard-to-navigate menus. Talking, it turns out, is far more intuitive than tapping through screens.
Seeing It in Action on the Show Floor
That vision came to life in a very tangible way at the Cisco booth at the National Retail Federation’s (NRF) Big Show this year.
Visitors were greeted by what appeared to be a Cisco employee standing ready to answer questions. They asked about the booth, the technology, and how retailers might use AI like this in a real store. The answers were immediate, conversational, and grounded in retail context.
Then came the second look.
The “person” was actually a hologram of Kaleigh, a real Cisco employee. The experience ran locally on Cisco Unified Edge with Intel Xeon 6 Processors and was powered by a retail-focused small language model (SLM) from Arcee AI. Instead of routing requests to a distant cloud service, inference happened at the edge; enabling fast, conversational responses without noticeable delay.
Under the hood, the architecture reflected how retailers could deploy similar capabilities in-store. Arcee’s SLM delivered store-specific intelligence with ultra-low latency and stable token streaming, supporting responsive, natural conversation rather than delayed fragmented responses. Cisco Unified Edge provided the infrastructure foundation delivering the local compute, networking, and secure management needed to run the model reliably at the edge. And Proto Hologram provided the immersive interface that made the experience intuitive and human.
The goal wasn’t to showcase a hologram for novelty’s sake. It was to demonstrate what becomes possible when AI runs at the edge. The same approach could support in-store assistants that help customers find products, suggest what they need for a specific project or recipe, troubleshoot issues, or guide them through complex decisions.
What Retailers Told Us
Conversations throughout the event reinforced a consistent theme: retailers are looking for AI that works in the real world, not just in demos.
Across roles and responsibilities, the questions tended to fall into two related camps. Teams responsible for IT and infrastructure wanted to understand how AI fits alongside the systems their stores already rely on; how it’s deployed, managed, secured, and kept reliable at scale. Business leaders and store operators focused on outcomes. They wanted to know what AI actually does on the store floor, how it helps short-staffed teams, and whether it simplifies or complicates day-to-day operations.
Both perspectives pointed to the same underlying needs.
Retailers don’t want to build everything themselves. They’re looking for integrated, turnkey experiences that can be deployed consistently across locations without custom integration work. Staffing shortages are real, and many newer employees don’t yet have the deep institutional knowledge customers expect. AI has the potential to act as a force multiplier, helping distribute expertise more evenly and supporting employees in moments that matter.
Language barriers also came up repeatedly, particularly for customer-facing use cases. Several retailers highlighted the importance of AI-driven experiences that can translate and respond naturally in multiple languages. That capability is quickly becoming a requirement, not a nice-to-have.
Just as important, retailers are cautious about AI becoming “another thing to fix.” Reliability matters. AI has to align with business KPIs and support existing store operations, not add fragility or overhead. Many teams emphasized the need for a platform that allows them to experiment to test new AI experiences safely, validate what works in real conditions, and scale those successes without disrupting critical applications.
Why Platform Thinking Matters at the Edge
Taken together, these insights point to a broader shift in how retailers think about edge infrastructure and who is expected to interact with it.
In most stores, the people closest to the technology aren’t IT professionals. They’re associates, managers, or regional teams who have to keep the store running. When something breaks or behaves unexpectedly, there often isn’t a dedicated expert on site to troubleshoot or intervene. That reality changes how edge infrastructure needs to be designed.
Supporting AI in the store isn’t just about powering a new experience. It’s about doing so in a way that minimizes operational burden from day one and throughout the life of the system. Retailers don’t have the luxury of standing up isolated environments, managing complex integrations, or relying on specialized skills at every location. Especially when stores are already running point-of-sale, inventory, security, and critical workflows.
That’s why platform approaches at the edge are becoming essential. Rather than treating AI as a bolt-on, retailers need a foundation that is simple to deploy on Day 0, easy to operate on Day 1 and resilient through Day N; all without requiring constant hands-on intervention.
This is where Cisco Unified Edge fits into the picture. Designed for distributed environments like retail, it brings together compute, networking, security, and cloud-based management into a single, modular platform. That allows retailers to evolve their in-store experiences over time without fragmenting their infrastructure or increasing operational complexity.
Just as importantly, a unified platform gives retailers room to experiment safely. Teams can test new AI use cases, validate what works in real store conditions, and scale confidently all while keeping critical applications stable, secure and easy to operate.
From Planning to Participation
For years, much of the retail AI conversation centered on planning: roadmaps, pilots, and proofs of concept.
That’s changing.
Retailers are no longer asking whether AI belongs in the store. They’re asking how to deploy it in ways that are practical, reliable, and aligned with the realities of running a retail business. Increasingly, the answer points to the edge.
The hologram wasn’t just a booth demo. It was a signal that retail AI is shifting from planning to participation and that the store has become the new edge.
If you’re looking to take the next step, we’ve developed industry-specific at-a-glances (AAGs) that outline practical deployment models for retail and other distributed environments:
