UCP-aligned reference · loyalty extension v2026-08-25 Synthetic demo merchant

Your loyalty program should still count when an AI does the shopping.

StabileRewards expresses your Shopify program's public terms, and a shopper's member value once they consent, in the loyalty format of the Universal Commerce Protocol, an open standard for AI-assisted shopping. Shopify stays your checkout and system of record.

What breaks

AI shopping compares price and shipping. Loyalty value goes missing.

When a shopper asks an assistant to find and compare products, the answer is built from catalog data: price, availability, delivery. A returning customer's points, their next reward, or what this cart would earn are usually not part of that comparison, even though they are real value you already pay for.

Invisible at the decision

The program exists on your storefront and in email, but not where the shopper is now deciding.

Compared on price alone

Without the loyalty value, you are compared only on the numbers any competitor can match.

Private data needs consent

Member balances and rewards must not leak to an assistant just because it asked.

What StabileRewards does

Public terms for everyone. Member value only with consent.

Anyone can see your program and what this cart would earn. A shopper's tier and benefits appear only after they authorize it. Earned rewards become ordinary Shopify discount codes, settled when the order is placed.

What StabileRewards does not do

A narrow loyalty layer, not another checkout.

  • Search or rank your catalog
  • Take payment
  • Place or change orders
  • Replace Shopify as system of record
  • Share balances or codes without consent
  • Promise placement in any AI assistant
What you can inspect today

One synthetic store, two carts, one member.

Harbor & Pine Tea is an invented Shopify merchant. Its numbers come from the same program logic StabileRewards serves to shopping agents today. Switch carts to see the Stabile result and its UCP loyalty representation side by side.

Synthetic data
Harbor & Pine Rewards

Loading program…

Scenario
UCP loyalty representation
…

Privacy boundary

Public
  • Program name and earn rate
  • Minimum spend and bonuses
  • What this cart would earn
Shopper-authorized
  • Member tier and benefits
  • Masked member id (****4821)
  • Earning without the first-order assumption
Stabile only, not UCP
  • Available points
  • Reward codes and point holds
  • Settlement after the order

Evidence

  • Published
    Fixtures and field mapping

    Three versioned JSON fixtures, each checked against the released UCP loyalty schema and a privacy denylist. Download from the brief.

  • Published
    Implementation notes for UCP contributors

    Four interoperability observations from mapping a Shopify points program onto the extension. Read the observations.

  • Live
    Anonymous loyalty quote

    The public, merchant-scoped quote endpoint this reference is derived from is live for merchants who enable it.

  • Live
    Same loyalty logic in ChatGPT, via MCP (not UCP)

    Our ChatGPT plugin is published in ChatGPT's app directory and uses the same program and reward logic through our MCP server. The plugin is an MCP integration, not a UCP one, so it is not evidence of UCP adoption by any platform. View the ChatGPT plugin.

  • Preview
    Walkthrough

    A 75-second timed walkthrough of the public quote, the consent boundary, a member benefit, and settlement, all on synthetic data. Play the walkthrough.

Next step

Request a pilot

We are recruiting a small group of Shopify merchants to put their real program in front of AI shopping flows, with their approval at every step.

A good fit if you

  • sell on Shopify and run, or want to run, a points program;
  • see shoppers arriving from AI assistants, or expect to;
  • want to approve exactly what is shown publicly.

Platform or UCP implementer? Use the same form and pick your role. We are glad to do a 30-minute interoperability review.

Sent to the founder's inbox via Formspree. Contact details are never sent to analytics. Privacy.