The now, next, and near-term applications of AI in retail

Retail is a naturally fast-moving sector. Shopper preferences can change in the blink of an eye, and so retailers are well accustomed to adaptation on the fly. But artificial intelligence (AI) represents a different kind of change – a change that’s now reshaping retail at every level, from pricing and promotions through to store assortments and media targeting.

We’re not talking near-time possibilities here, either. We’re talking about real impact from projects that are already underway. Right now, for instance:

• 78% of global organisations are using AI in at least one business function¹
• 92% of retail marketers are actively using AI²
• 76% of retailers say they plan to deepen their investments in AI over the coming year³.

Clearly, there’s no shortage of excitement here – but just where is that money going? Where is AI already having an impact? What’s just around the corner? And what could retail look like when truly intelligent systems can act on our behalf?

Let’s take a look at the today, tomorrow, and the mid-term future of AI in retail – and what it all means for retailers and customers alike.

Quietly transformational: AI in retail today

While generative AI might be dominating the headlines right now, the most impactful uses of AI in retail are already embedded behind the scenes – quietly driving efficiency, insight, and customer relevance at scale. These use more traditional (but no less powerful) AI methods such as machine learning or optimisation.

Take pricing and promotions. Balancing business rules, supplier agreements, competition, and customer perception effectively is a task that now goes beyond manual capabilities. With AI, retailers can simulate thousands of scenarios and identify the sweet spot between revenue, profit, and loyalty. Retailers working with dunnhumby, for example, have seen 1–3% gains in comparable sales using AI-powered pricing strategies.

AI is having a similarly additive impact around personalisation and retail media, too. From recommending your favourite coffee just before you run out, to tailoring offers in real time via in-store apps or social channels, omnichannel personalisation is no longer aspirational – it’s essential. Take the results of one recent Nestlé campaign at Tesco: here, dunnhumby’s AI-led audience targeting led to an 11% uplift in sales.

And then there’s assortment. Ranges that used to be set at chain or regional level are now becoming hyperlocal. By analysing shopping behaviours and store-level patterns, AI enables more relevant product ranges, reduces waste, and improves availability.

These are mature, proven applications delivering measurable value right now. But what’s next?

GenAI gets going: near future applications

We’re rapidly entering a second wave of AI adoption – this time powered by generative AI. While the technology is still maturing, its early impact is already being felt in two key areas: productivity and customer experience.

Let’s start with one of the clearest applications; used effectively, GenAI can augment human capabilities. Tools like GitHub Copilot and other AI coding assistants can help data scientists document workflows, debug code, and accelerate delivery, freeing up time to focus on solving bigger, more strategic problems. For marketing and ecommerce teams, meanwhile, GenAI can generate product descriptions, localised campaign copy, and social content at scale.

On the customer side, GenAI is powering smarter, more responsive service. Many retailers have rolled out chatbots to help customers search for products, navigate offers, and even receive recipe suggestions in real time.

That said, the road to widespread adoption isn’t without obstacles. Today, issues like model hallucinations, inconsistent output, and lack of domain-specific knowledge still hold GenAI back from playing a role in truly business-critical applications, along with more industry-wide challenges like data privacy, bias, or regulatory concerns. But it’s in this area in which the longer-term opportunity lies.

Agents, autonomy, and machine shoppers: near-term use cases

The next frontier in retail AI – and it’s coming faster than you might imagine – is what’s being called agentic AI: systems that don’t just assist people, but make decisions and take actions on their own. Think about it this way: today, most AI tools wait for a person to prompt them with a specific question or task. Agentic AI, by contrast, won’t need that prompt. It will plan and act independently, based on the signals it receives and the goals set in advance, without the need for constant human interaction.

In practice, that could mean prices adjusting automatically as demand shifts, inventories that manage themselves, or store layouts redesigned and tested by AI without weeks of manual work. Instead of lengthy processes that require multiple teams to hand tasks off to one another, agentic AI can keep things moving by making optimal decisions in real time – always aligned to the objectives set by people. The result: operations that are faster, more efficient, and far more responsive to consumer needs.

While still emerging, the momentum is clear. According to Gartner, only 1% of enterprise software today uses agentic AI, but by 2028, that number is expected to rise to 33%⁴. To unlock the potential, retailers need orchestrated systems where multiple AI ‘agents’ communicate with one another, each responsible for a specific part of the retail value chain: pricing, assortment, media, supply chain, and more.

This shift also raises new questions about brand loyalty and customer experience – especially as we approach the era of “machine consumers.” Gartner predicts that by 2030, 20% of retail revenue will come from AI agents shopping on behalf of humans. That means retailers and brands will need to influence not just individuals, but the algorithms making choices for them – ensuring their stores and products are chosen when decisions are driven by data, logic, and personalised preferences.

Change is already happening

If there’s one message to take from this three-phase view of retail AI, it’s this: transformation isn’t coming – it’s already happening. The real challenge now is how to scale what works. Retailers and brands need to cut through the hype, and prepare for a future where intelligent systems aren’t just tools, but collaborators – and even decision-makers as well.

As with any sector, retailers need context-aware AI systems that understand their data, operations, and decision-making logic. This is where dunnhumby enters the equation: as well as our own AI-powered tools, we can also help retailers fine-tune AI systems with curated retail data, structured hierarchies, and expert input to ensure outputs are accurate, explainable, and truly useful.

Whether it’s helping optimise current operations, guiding the adoption of GenAI, or shaping the agent-led retail ecosystems of tomorrow, we’re here to help retailers make every decision more customer-focused, efficient, and future-ready.

¹ The state of AI: How organizations are rewiring to capture value – McKinsey, March 2025
² Over 40% of agentic AI projects will be scrapped by 2027, Gartner says – Reuters, 25^th June 2025
³ Machine Customers—AI Buyers To Control $30 Trillion In Purchases By 2030 – Forbes, 18^th February 2025
⁴ Intelligent Agents in AI Really Can Work Alone. Here’s How – Gartner, 1^st October 2024