If a Customer Shops Through AI, Can It See What They've Already Earned?
Direct Answer AI shopping assistants are starting to decide where a customer buys, but most of them still cannot see what that customer has already earned with a merchant. A $5 price gap can flip once an assistant accounts for an available reward — but only if it knows the reward exists. Stabile Loyalty Rewards Agent brings that value into the conversation itself: check a balance, see what a cart would earn, redeem what's already there. Version 2.0 — a five-tool surface with real program data, not sample numbers — published in the ChatGPT plugin directory on September 9, 2026.
If a customer is shopping through AI, can the AI see and use the rewards that customer has already earned?
Most loyalty programs were not built to answer that question, because they were not built for this shopping journey.
The journey loyalty was designed for is not the one happening now
A customer visits a merchant's site or app, signs in, finds the points balance, opens the rewards page, checks the rules, and — if all of that happens in the right order — remembers to use a reward before checkout.
That model works when the customer is navigating the merchant's own experience.
AI shopping increasingly starts somewhere else entirely:
"Find me a good pair of running shoes under $150."
"Which store gives me the best value?"
"Can I use any rewards on this?"
The customer may never open the merchant's rewards page at all. The assistant is standing between the customer and that page, and it decides what gets surfaced.
That is a new channel, and most loyalty programs are not in it.
Price alone does not answer the question the assistant is being asked
AI shopping systems are getting better at reading catalogs, prices, reviews, and policies. That is necessary. It is not sufficient, because a customer's relationship with a merchant carries value the catalog does not contain — points, a standing discount, free shipping, a status tier, cashback, a reward sitting unused.
Two merchants selling the same product make the point concrete.
Merchant A: $80. Merchant B: $85.
Read as list prices, Merchant A wins.
Now suppose the customer already holds a $15 reward with Merchant B.
- Merchant A: $80
- Merchant B: $85 − $15 = $70
The better deal changes entirely, but only if the assistant can see the reward. If it cannot, the loyalty program has no vote in an AI-mediated purchase — not because the program is weak, but because it is invisible to the thing making the recommendation.
From a page customers have to remember to a capability the assistant already has
Loyalty has mostly been built as a destination: visit our rewards page, open our app, log in to see your balance. That model assumes the customer will go looking.
AI shopping makes a different model possible — loyalty as something the assistant can reach for in the moment it matters, without the customer needing to know where the rewards page lives or that one exists.
The shift matters because a reward earns nothing merely by being issued. It earns something when it changes a decision. If customers increasingly let an assistant decide where to buy, a loyalty program that the assistant cannot see is a loyalty program that cannot do the one thing it exists to do.
In practice, that looks like ordinary conversation, not a new interface to learn:
"Do I have rewards at this store?"
"How many points will this purchase earn?"
"What can my points unlock?"
"Can I use a reward on this cart?"
"Redeem it for me."
None of it requires the customer to understand the merchant's loyalty stack. They just ask.
Where this actually stands today
Being specific about what is live and what is not, because the difference matters more here than almost anywhere else on this blog: a reward is real money, and the standard for what an assistant reports about it is "correct, or clearly unknown" — never a guess dressed up as a balance.
Live since August 25, 2026. Stabile: Loyalty Rewards Agent first published in the ChatGPT plugin directory as v1. Its tools return a merchant's actual program terms — earn rate, tiers, minimum spend — rather than fixed sample numbers. When a shopper cannot be identified, it says the balance could not be checked rather than reporting zero, so nobody is told they have no rewards when they do.
Proven end-to-end. The full redemption chain — quote, reserve, apply, settle, decrement, bill — has closed on a real order through a live agent connector. That is the hard part, and it is done.
Published today. Version 2.0 — five tools, a corrected privacy disclosure, and a customer-identity guard on redemption — was approved and published in the ChatGPT plugin directory on September 9, 2026. (OpenAI's own version counter runs slightly ahead of ours for build-level fixes, which is a bookkeeping detail, not a different release — publicly, this is v2.0.) We have written before about how long that review can take — ten submissions this round.
So: "check what I've earned" and "redeem it for me" are both live, published capabilities today, not a roadmap item. What's still ahead is the same thing that's always ahead of a capability that has proven itself once — watching it hold up at the volume of ordinary ChatGPT shopping traffic, not just a proof run. We'll report on that as it happens.
What merchants should start asking
The old question was "Do we have a loyalty program?"
The current one is "Can an AI shopping agent see what our loyalty program offers this customer?"
And past that: "Can the agent actually help the customer use it?"
These matter because loyalty is fundamentally about preference — a merchant gives customers a reason to come back, and that reason only works if it can reach the moment where the customer decides. If a new channel sits between the customer and the merchant, loyalty has to be visible inside that channel, not just at the destination the channel routes around.
The objective has not changed
Checking and redeeming rewards is a starting point, not the ceiling. The longer arc is an assistant that can reason about loyalty the way a person would — whether spending points now beats saving them, whether a reward changes which merchant actually offers the better deal, which available benefit matters most for a given purchase. That turns a balance into something closer to purchasing power.
What merchants have always wanted from loyalty has not changed: recognize the customers worth keeping, and give them a reason to choose you again. What changed is where that decision gets made.
Make the customer's existing relationship with you count when the next purchase decision happens — even when that decision happens inside a conversation you never see.
Get started
View Stabile ChatGPT Plugin listing ->
If you run a Shopify store and want your loyalty program visible to AI shopping agents:
Install StabileRewards from the Shopify App Store ->
Free to install · Pay on results: 1% of reward value redeemed + $0.05 per AI-channel transaction · No charge if no rewards are applied.
Questions about the v2 listing: get in touch or email support@stabilerewards.com.