Move over Generative AI, there’s a new smart technology in town. Yes, ChatGPT might not even be three years old yet. But in these accelerated digital times, a new pretender to the AI throne is already threatening to steal the limelight. It goes by the name of Agentic AI.
Agentic AI is essentially the next step in intelligent machine automation. By ‘agentic’, the name refers to the fact that this new generation of AI has ‘agency’, or the ability to make decisions independently. And not just the limited rule-based decisions that previous generations of automated systems could make. We’re talking decisions (and strings of decisions) that require the kind of higher-level reasoning and inference skills that humans use. Only, being software, AI agents are able to execute these decisions with much greater speed, scale and accuracy.
These advanced reasoning capabilities have arisen from the evolution of the Large Language Models (LLMs) that power Generative AI, while the whole idea of a helpful AI ‘agent’ continues the tradition of chatbots and smart assistants. But unlike Gen AI, which needs human prompts to keep generating outputs, Agentic AI can act autonomously, either taking over entire workflows independently or working as a collaborative ‘friend’ to human operators. And unlike the AI chatbots we’ve become used to so far, the key point about AI agents is that they can act as well as talk – they don’t just dispense advice, they do what needs doing.
Unsurprisingly, this kind of next-gen digital assistant is attracting a lot of attention from businesses seeking to optimise internal processes for efficiency and productivity. According to eMarketer, this includes more than half of retailers, who are already evaluating Agentic AI’s potential.
So what exactly does this new technology have to offer retail, and in-store vendors in particular?
A smarter retail experience
Every sector can benefit from improved operational efficiency, and retail is no exception. Robotic Process Automation (RPA), or the use of specialised software ‘bots’ to automate repetitive digital tasks, has been successfully used by retail operations to lighten the load of time-consuming processes like inventory checks or pricing updates.
But RPA is largely a back-office capability. An exciting prospect with Agentic AI is the promise of having it embedded in front-line workflows, straight through POS systems, and handling the likes of inventory management, pricing adjustments, and transaction processing right off the bat with minimal instruction.
Not only would this reduce manual workload, it would also allow decisions to be executed in real time throughout the store based on transactional data. An obvious example would be real-time stock updates being displayed on digital signage on the shelves, or replenishment alerts appearing on staff devices. A more advanced example would be Agentic AI opening the door to full-blown dynamic pricing, or price adjustments being made in response to fluctuating demand to achieve the optimum balance of sales volume and margin.
The idea of AI agents controlling aspects of what customers see via digital signage reveals another huge area of potential in retail – the ability of Agentic AI to shape the customer experience. Amazon Web Services (AWS) recently published an article setting out a vision for a futuristic ‘Agentic Store’, where a combination of sensors, scanners, and signage allow advanced AI tools to orchestrate every aspect of the in-store experience, from stock-out warnings to organising alternative payment arrangements if a customer’s card doesn’t work.
The journey to the Agentic Store might start with AI agents being deployed on kiosks to refine the self-service process further still, combining highly efficient process automation with personalised suggestions and recommendations. Unlike the current frustrating experience of having to call for a human assistant when an item doesn’t scan at self-checkout, Agentic AI tools could problem solve the situation themselves. And there are also hopes that the technology can help to cut down on theft by spotting suspicious behaviour patterns.
Longer term, personalisation stands out as perhaps the biggest opportunity Agentic AI represents in retail. Whether via facial recognition technology or picking up the digital fingerprints of customers’ devices, agentic systems could identify returning customers and offer tailored recommendations and discounts – perhaps even via their own devices as they shop. It’s a prospect that certainly seems to chime with consumers. As per eMarketer again, 70% of people are already interested in AI agents helping them manage and optimise loyalty points and rewards.
There’s still a way to go before we reach such a future, and critical issues to resolve around things like data governance. But given the pace of change we’ve already become accustomed to with AI, it makes sense to start considering the possibilities of an agentic future, today.