Most brands think the biggest threat to their loyalty program is a competitor with a better points program. The bigger disruption, however, is structural, and it’s already live. Agentic commerce, where AI agents shop, compare, and transact on a customer’s behalf, is rewriting who controls the moment of purchase. And here’s the rub: most loyalty programs weren’t designed for a world where the shopper isn’t human.

In late 2025, OpenAI launched Instant Checkout in the US, enabling ChatGPT users to buy products directly in conversation via the Agentic Commerce Protocol (ACP)¹. By January 2026, Google’s Universal Commerce Protocol (UCP) debuted at NRF with early retail and platform partners², including Walmart, Target and Shopify. Adobe reported AI-driven traffic to US retail sites surged 4,700% year-over-year in mid-2025³.

While most of these implementations remain early, constrained and unevenly deployed, the direction of travel is unmistakable. This is no longer a theoretical shift to observe. It is an infrastructure transition that brands must actively prepare for.

This blog sets out to cut through the noise: what agentic commerce demands of loyalty programs, why the stakes are higher than most brands currently appreciate, and what it takes to compete when the ‘person’ doing the shopping is an AI agent.

The shopper has changed…… again

Retail has always rewarded the brands that moved first. One-click checkout, mobile-first commerce, buy now, pay later. Each shift sorted the fast movers from the late adopters. Agentic commerce is the next inflection point, except the pace of infrastructure build-out is unlike anything we’ve seen before.

I like to think of it this way: a human shopper can be persuadable at the point of contact. They respond to well-timed creative, a compelling email, a meaningful in-store moment. An AI agent operates on an entirely different logic. It doesn’t get distracted by a homepage banner. It was given a brief before it started – a set of parameters the customer defined in advance, and it works through that brief methodically. Price ceiling. Preferred brands. Delivery requirements. Sustainability filters. The agent isn’t browsing. It’s executing.

Two US examples illustrate how fast this is moving. Walmart and Sam’s Club went live on Google Gemini in January 2026, letting customers shop in natural language via UCP – describing what they need and having the agent handle product matching, availability checks, and checkout. Instacart became the first grocery partner for ChatGPT Instant Checkout in December 2025, enabling users to order without leaving the conversation. Modern Retail puts daily shopping-related queries on ChatGPT at around 50 million⁴. While most of these interactions are still informational rather than transactional, the sheer volume of commerce-oriented intent flowing through conversational interfaces is already commercially meaningful. 

What this means for brands is straightforward but significant: the channel has changed and so has the currency of influence. Creative and content still matter, but upstream. At the point of agent decision, what matters is whether your brand’s value can be read, verified, and acted on by a machine.

What does loyalty mean when an agent is working on your behalf?

Here’s the question brands should be contemplating right now: if a customer’s AI agent is shopping on their behalf, what would make it choose you? Not what would make the customer choose you – they’ve already made decisions about their preferences before the agent started. What would make the agent execute in your favour?

The answer is that loyalty must perform three jobs it was never explicitly designed for – and it must perform them in real time, across every channel, without fail.

The first job is authentication. An agent can only act on verified information. It can’t infer that a shopper is a Gold tier member with 15,200 points available – it needs that data served to it cleanly, via a connected system, the moment it’s relevant. If your loyalty program lives in a silo that doesn’t talk to the channels an agent is working through, your member benefits effectively don’t exist at the point of decision. That’s a lost sale.

The second job is personalisation fuel. The data a loyalty program holds (first party) – what customers buy, how often, in what categories, at what price points, is the exact input that makes AI-driven personalisation useful. Scot Wingo, founder of ReFiBuy and author of the Retailgentic Substack, has written at length on how AI-driven decision systems can match shopper context to the right offer better than any previous technology – but only if the underlying data is structured, permissioned, and accessible. Loyalty programs that collect this data well are sitting on a strategic asset most brands are significantly underutilising.

The third job is differentiation that survives price comparison. When an agent is evaluating options, commoditised benefits – basic discounts, standard free shipping – don’t tip the scale. What does tip it is value that is genuinely hard to replicate: access to exclusive product drops, status recognition that changes how a customer is served, partner benefits that extend the program’s utility beyond a single brand. These aren’t just ‘nice loyalty features.’ In an agent-evaluated world, they are the reason a customer’s parameters include your brand in the first place⁵. And only a well-designed loyalty program can hope to have these in the first place.

The loyalty marketer’s expanded job description

Most loyalty managers today are measured on metrics driven by the human experience: member acquisition, activity rate, reward redemptions, NPS among program members. Those metrics still matter. But a new set of operational questions is emerging alongside them – and right now, most loyalty teams aren’t the ones asking them.

Is your program’s eligibility logic available via API? Can an external system, an AI agent, a third-party platform or a voice interface query your member benefits in real time and get a reliable, consistent answer? Are your earn and redeem rules the same across your website, your app, your POS, and across any agent surface a customer might use? These questions used to sit with the technology team. In an agentic commerce world, they are loyalty strategy questions – and the loyalty manager is the person best placed to own them.

There is a deeper strategic risk that brands must also confront. As agents become the primary interface to commerce, they will not simply consume loyalty benefits, they will increasingly abstract and normalise them.

Agents will optimize across merchants, surface equivalent benefits from different programs, and potentially aggregate rewards, credits and entitlements at the agent level rather than at the brand level. Over time, this creates a real possibility that the customer’s perceived loyalty shifts away from individual brands and toward the agent that is orchestrating value on their behalf.

In this environment, loyalty programs that rely primarily on interchangeable mechanics such as points per dollar, generic discounts and standard free shipping, become especially vulnerable. If the benefit can be normalised, it can be substituted.

The strategic challenge for brands is therefore not only how to integrate their loyalty program into agent ecosystems, but how to design benefits and entitlements that resist abstraction. Loyalty must create advantages that an agent can recognise but cannot easily flatten into a generic optimization layer.

Think of it like building a house. The interior design (e.g., what the program feels like to a member), how it’s communicated, what it rewards, etc, is the loyalty manager’s traditional domain. Agentic commerce adds a new requirement: the plumbing and wiring must work perfectly, because now external systems need to connect to those pipes. A beautifully designed program with broken backend logic won’t get selected by an agent. It’ll just get skipped.

This is a genuine expansion of scope. It means loyalty managers need to be in the room when API architecture decisions are made, when data strategy is set, and when new commerce channels are evaluated. The program brief must include how the program performs for a machine, not just how it feels for a human.

How should brands prepare for agentic commerce?

The underlying principles and best practices of loyalty haven’t changed. What has changed is the cost of getting them wrong. Here’s where the focus should be. 

Unify your incentive data before integrating anything

Promotions in one system, member pricing in another, tier benefits somewhere else – that architecture was manageable when the shopper was a person who could navigate inconsistency. An AI agent can’t. 

It needs a consistent answer every time it queries your program, regardless of which channel the transaction happens in. Getting to a unified incentive layer isn’t just a technology project. It’s a prerequisite for agentic commerce participation.

Translate your value proposition into something a system can read

The direction both ACP and UCP are moving is clear. Loyalty benefits need to be expressible as structured data: specific pricing by product, threshold-based shipping by tier and redeemable credits mapped to checkout events.

But in practice, this will extend beyond simple data exposure. Loyalty programs will increasingly need to express their rules, eligibility logic and entitlements as machine-interpretable policies. Effectively, these are executable loyalty contracts that an agent can evaluate, validate and apply without human interpretation.

If your program’s value only exists in a PDF, a set of creative assets or a marketing headline, it is basically invisible to an agent. This isn’t about rebuilding your program from scratch. It’s about ensuring that what you have built can be interpreted and applied by the systems that will increasingly complete purchase decisions on your customers’ behalf.

Build benefits that raise the floor, not just the ceiling

As alluded to above, loyalty programs that rely on transactional incentives, such as points per dollar, a birthday discount, or a free shipping threshold, are building on a floor that agents can easily level. Every competitor can match those mechanics. 

The programs that hold up in agent-evaluated environments are the ones with benefits that change the nature of the relationship: faster service, priority inventory access, experiences that aren’t available anywhere else. These raise the floor of what it means to be a member, rather than just adding points to the ceiling.

Think about where loyalty sits in the customer’s agent instructions

Ultimately, the goal of agentic commerce loyalty isn’t just to be present on agent platforms, it’s to be present in the parameters a customer sets before the agent starts. That happens when a program is genuinely valued enough that a member thinks: ‘I want my agent to factor this in.’ 

Getting there requires earning that trust through consistent delivery, not just signing up to an API standard. This goes back to the consistent theme of building trust with loyalty program members across their entire experience with the brand. 

Summary

Agentic commerce isn’t arriving. It’s already well and truly here. And the brands best positioned to compete in it aren’t necessarily the ones with the biggest loyalty programs, they’re the ones with the most functional ones.

The programs that will struggle are the ones built purely for human shoppers, with fragmented data, generic benefits, and back-end logic that doesn’t talk to external systems. The programs that will hold up are the ones that can authenticate a member in real time, surface their entitlements accurately, and offer advantages that cannot be flattened or substituted by an agent’s optimization layer.

Loyalty was always about building a relationship architecture worth coming back to. Agentic commerce adds a new test: is that relationship legible enough that a machine would recommend it? When brands get both of those right, the program stops being a retention tactic and starts being a genuine source of commercial advantage.

__________________________________________________________________________________

Max Savransky is the Global Director of Loyalty Strategy at TrueLoyal. Max is a customer strategy, loyalty and data leader, with a proven 17-year track record of designing, validating and deploying successful client strategies to drive engagement, retention and revenue growth. Max is also one of the co-authors of ‘Loyalty Programs The Complete Guide’ (editions 1 and 2), the definitive book on loyalty for industry professionals.

Connect with me on LinkedIn!

__________________________________________________________________________________

Want to find out more about Agentic Commerce and what it means for your loyalty program? If you’d like to chat about how we can help, get in touch with our Enterprise Sales team 

__________________________________________________________________________________

 ¹OpenAI (2025). Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol. Referenced in: intro (ACP launch, Instacart partnership); H2 #4 (ACP data standards).

 ²Google / NRF (January 2026). Universal Commerce Protocol announcement and retail partner launch. Referenced in: intro (UCP launch); H2 #1 (Walmart/Sam’s Club Gemini integration); H2 #4 (UCP data standards).

³Adobe Digital Economy Index (2025). AI-driven retail traffic and Cyber Monday 2025 analysis. Referenced in: intro (4,700% AI traffic growth statistic).

⁴ Modern Retail (January 2026). Why the AI shopping agent wars will heat up in 2026. Referenced in: H2 #1 (50 million daily ChatGPT shopping queries).

⁵Scot Wingo, Retailgentic Substack (2025-2026). Retailgentic – AI, agentic commerce and the future of retail strategy. Referenced in: H2 #2 (AI personalisation and data structuring); H2 #2 (differentiated benefit design).