When the Shopper is a Machine: The Loyalty Implications of Agentic Commerce

Anyone who operates a loyalty program knows that they’re designed to influence buying behavior ...  human buying behavior. A customer remembers their points balance, activates an offer, and checks the app before checkout. That model is showing its age.

Artificial intelligence agents are actively making real purchase decisions on behalf of consumers, both within the third-party AI platforms consumers routinely use for everyday queries as well as in retailer-owned chatbots within their own e-commerce channels. Loblaw launched a PC Express app inside ChatGPT that lets customers plan meals and push ingredients to their basket. Albertsons and Target are testing conversational advertising through ChatGPT, embedding agentic AI tools across the customer journey. Woolworths, on the other hand, is building Olive, a Google Gemini-powered shopping companion that can plan meals, interpret handwritten recipes, and construct baskets without human intervention, but within the Woolworths shopping environment.

These pilots all point to retailers’ reckoning with the shift in shopping behaviors. But when an AI agent decides where to shop, what happens to loyalty programs ... and a brand’s relationship with the customer?

Solving the Speed Problem of AI Acceleration

Batch processing is one element of traditional loyalty operations that will hold programs back in the agentic age. Weekly refreshes, overnight syncs and segment-based personalization updated every few days worked when customers had time to browse, compare and remember. AI agents operate differently.

For those getting up to speed in agentic commerce speak, an AI agent calculates the optimal value for every available program by evaluating all the data shared by retailers through their APIs as well as the user’s criteria. And it does this in less than an instant.

Whether it’s a third-party AI agent or a shopping assistant created by a retailer, this is already happening. When an AI agent reviews a shopping cart, for example, it queries tier status, point balances and offers eligibility in milliseconds. In the case of Olive, if a customer is $5 short of a "Spend $50, Get $5 Off" threshold, the assistant identifies that gap and suggests an add-on to capture the discount. The agent makes this calculation before the customer reaches checkout.

If your loyalty engine can’t surface a similar verified offer during the agent's API call, the agent moves on and defaults to the lowest shelf price elsewhere. Your program becomes invisible, and that’s a real danger.

3 Requirements for Discovery and Checkout

Instantaneous response times are just one of the three capabilities loyalty programs must deliver to remain part of AI-driven purchase decisions.

1. Machine-readable rules: Agents parse structured data, not marketing copy. Your offer terms, eligibility criteria and redemption conditions must be exposed in formats agents can evaluate without interpretation.
2. Sub-second response times: When an agent makes a discovery and queries your platform during peak traffic, you need to return validated offer data in under 150 milliseconds. One or two seconds might as well be one or two weeks.
3. Real-time validation and redemption: Agents need to confirm that the offer is valid for this customer, at that moment and apply it at checkout. Any gap between validation and redemption is an opening competitors can exploit. Today, for most AI agent experiences, the checkout happens outside the agent flow and redirects back to traditional e-commerce pages. But it doesn’t change anything about the requirements for your platform to be able to support real-time validation and redemption: different surface for the customer, same backend that serves offers and promotions.

Adapting Your Platform for the Agentic Age

If your loyalty program processes offers overnight or relies on "near real-time" syncs that take seconds or minutes, it’s built for a different era. Agentic commerce demands infrastructure that can issue offers, confirm status, and redeem discounts on timescales measured in milliseconds, and that should be your top priority.

If your tech stack can’t validate and apply a personalized offer in under 150 milliseconds while handling five times your usual Black Friday traffic, your work starts now.

And it should start now. There are already retailers deploying AI shopping assistants today, and they’re laying the groundwork for a model that will soon be standard. Don’t let the hard work you’ve put into your loyalty strategy over the years be swept away by the agentic wave overnight.

Jean-Matthieu is the Eagle Eye Group’s first Chief AI Officer, bringing his pioneering, forward-thinking AI expertise to retail solutions.