AI agents are already making purchase decisions. Loyalty programs without real-time infrastructure won’t be part of the calculation
The way we buy groceries is starting to change, and most loyalty programs aren’t ready for it.
In January, Woolworths became the first Australian retailer to adopt Google’s Gemini Enterprise for Customer Experience platform, announcing it will upgrade its Olive chatbot from a customer service tool into an AI shopping agent capable of planning meals, interpreting handwritten recipes, and building baskets on a shopper’s behalf.
Weeks later, Canadian grocery giant Loblaw launched its PC Express shopping app inside ChatGPT, allowing shoppers to explore meal ideas and curate ingredient lists through OpenAI’s chatbot before purchasing from their local store.
These are early moves in what will be a structural shift in retail. AI-powered agents are beginning to sit between the shopper and the checkout, making product selections, comparing prices, and identifying promotional value on the customer’s behalf.
Traditional loyalty often relied on customer memory; you’ve got to remember to scan your card, remember your points balance and remember to activate certain offers to maximise value. This creates barriers to redemption, protecting program economics through breakage or lack of awareness.
AI agents eliminate this entirely. They don’t forget. They don’t need to check an app. They calculate optimal value across every available program instantly, at the moment of decision.
The Woolworths and Google announcement at NRF illustrated how this plays out in practice. The companies shared how Olive will act as a fiduciary for the shopper. When a shopping agent reviews a cart, it can query tier status, point balances, and offers eligibility before checkout, all in milliseconds. If a customer is just shy of a threshold, say, spending $45 when a “Spend $50, Get $5 Off” offer is active, Olive can identify that gap instantly and suggests a $5 add-on to secure the discount.
If your loyalty engine can’t verify that offer during the agent’s API call, the agent will optimise for the lowest shelf price elsewhere. If your platform can’t adjudicate in that window, your program is invisible to the agent.
Why Most “Personalised” Programs Will Fail
The word “personalisation” is widely used in loyalty marketing, but the infrastructure behind it varies considerably. Most loyalty platforms claiming to offer personalisation are actually running sophisticated segmentation on batch cycles, with weekly refreshes, overnight processing, or offers that update every few days.
That worked when humans were making decisions. It is completely inadequate when machines are. Google’s roadmap highlights a shift toward Agentic Consent, where customers give agents permission to access member-only pricing and offers automatically. This creates a high-stakes environment: if your platform lags while verifying a member ID, the agent may default to a competitor whose API responds faster.
AI agents have straightforward but demanding requirements from loyalty infrastructure. They need simplicity, meaning machine-readable rules with no hidden terms. They need speed, specifically sub-second response times during peak traffic. And they need instantaneous accuracy, covering real-time balance verification and offer adjudication.
How fast a brand’s loyalty and personalisation offering is will have a big impact.
Marketing Technology News: MarTech Interview With Fredrik Skantze, CEO and Co-founder of Funnel
Built for Speed
At Eagle Eye, we have been focused on real-time performance because we saw this shift coming. Our real-time platform, powered by Google Cloud, can handle the issuance and redemption of thousands of completely personalised offers per second, across all channels, including in-store.
That architecture serves two purposes simultaneously. The same infrastructure that enables real-time personalisation for human shoppers is exactly what AI agents like Olive will query. The difference is that agents will demand it every single time, not only during promotion windows.
In practical terms, the platform is built to handle two specific requirements. Real-time offer issuance based on live customer behaviour and context, with 1:1 personalisation at an individual customer level, not segment pools. Real-time offer redemption at checkout, online and in-store, with sub-250ms response times at true peak loads.
The Question Every Loyalty Leader Should Ask
As AI agents start making more purchase decisions, loyalty programs will increasingly be calculated rather than felt. The programs that win will be visible, fast, and easy for machines to understand.
Two questions are worth putting directly to your technical team. On issuance: can your system detect a customer’s unique context, such as their current location, the weather, or the specific items just added to their digital basket, and issue a 1:1 personalised offer in that moment?
On redemption: can your system validate and apply a personalised discount, such as a “$5 off $50” offer available only to a subset of customers, in under 250 milliseconds while handling five times your usual peak traffic?
The answers will tell you whether your platform is positioned for the current environment or the previous one. If your platform processes offers overnight or relies on near real-time syncs that take seconds or minutes, it is built for yesterday’s world. In agentic commerce, a 10-second delay might as well be a 10-day delay. The agent has already moved on.
The foundation for agentic commerce is real-time personalisation, not as a feature, but as infrastructure.
Marketing Technology News: The Death of Third-Party Cookies Was Just the Start. Are You Ready for Consent Orchestration?
