Mastercard's Proto sandbox tests if AI agents find you
Mastercard is launching Proto in the UK in August so retailers can test agent discoverability and agentic chargebacks before real money moves.
A card network building a test rig is a useful signal about where risk is moving. Mastercard is launching Proto, an agentic AI sandbox, in the UK next month — and the first thing it lets retailers check is whether AI shopping agents can find their products at all. When Mastercard starts treating agent discoverability as something you measure rather than assume, it's stopped being a marketing topic.
What actually happened
Mastercard picked the UK as the first market for Proto, an expansion of the Agent Suite it launched in January. Per Finextra and FStech, Proto goes live in August 2026 and lets retailers and their banking partners test three things:
- Whether their products are discoverable by AI agents
- How to scale agentic payment options without losing consumer trust
- Whether their dispute handling holds up when agentic sales hit volume
Mastercard is also shipping three agents into the suite — a shopping agent, an onboarding agent, and a dispute agent. Simon Forbes, Mastercard's UK and Ireland division president, framed the UK choice around its retail sector and fintech scene.
Why agent discoverability matters for your business
Point one is the one to act on today, and it doesn't require a sandbox or a UK entity. Agent discoverability is a property of your product data — structured attributes, machine-readable inventory, a feed an agent can parse without guessing. If a shopping agent can't tell what size, material, or availability your SKU has, it recommends the competitor whose feed says so plainly. You don't get a bounce report for that. You just don't appear.
Point three is the one nobody has budgeted for. A chargeback today assumes a human who can be asked what they intended to buy. When an agent placed the order, "I didn't authorize this" gets murky fast — was it the agent's inference, your product data, or the shopper's instruction? Mastercard building a dispute agent and a dispute test bench in the same release tells you they expect the volume to be real.
The practical read: treat agent discoverability as a testable property of your catalog, not a channel you sign up for. Audit your feed the way you'd audit a public API. Then write down, before it happens, who eats the loss when an agent buys the wrong thing.
Key takeaways
- Mastercard's Proto sandbox launches in the UK in August 2026, expanding the Agent Suite it introduced in January
- It tests agent discoverability, agentic payment scaling, and dispute handling at volume
- Three new agents ship with it: shopping, onboarding, and dispute
- Discoverability is a product-data problem you can fix now without waiting for a sandbox
- Agentic chargebacks are a policy gap — decide who owns the loss before the first one lands
Most catalogs fail the agent test on attributes, not on tech. We audit product data for machine readability and build the feeds agents can actually parse. Send us your catalog and we'll tell you what's missing.
- #agentic-commerce
- #mastercard
- #ai-shopping-agents
- #chargebacks
- #product-data
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Agentic AI spending: $944B now, $3.35T by 2030 — read it right
A new PHD/WARC forecast puts agent-facilitated consumer spending at $944B in 2026, rising to $3.35T by 2030. That's 3.8% of consumer spend. Here's the operator read.
Read itInstacart buys Arpalus: who owns your shelf data?
Instacart acquired computer-vision startup Arpalus to turn 600,000 shoppers into a shelf-intelligence sensor network. If they run in your stores, they know your inventory better than you do.
Read it