top of page

AI Agents Can Now Read Card Rewards: What Imprint's $130M Raise Means for Brands

3 days ago
5 min read

An AI shopping agent may soon do more than find a product and fill a basket. It could also check which of a shopper’s cards earns the best reward, select a payment option and initiate a purchase, subject to the customer’s approval.

That is the commercial significance of Imprint’s latest announcement. On 11 November 2026, the co-branded card platform said it had raised $130 million at an approximately $1.9 billion valuation, in a round led by Khosla Ventures. Hedosophia, Ribbit Capital, Spice and Thrive Capital also participated, alongside new investor Positive Sum. The valuation is up from $1.2 billion at Imprint’s $150 million Series D in December 2025, according to VCPost.

Alongside the funding, Imprint released a Model Context Protocol (MCP) server designed to let AI agents access cardholders’ rewards, virtual cards and transaction history. It also introduced Imprint Access, a refundable deposit backed card for eligible people who are new to credit, with a pathway to an unsecured card.

The funding matters. But for brands, the larger strategic question is who owns the moment of payment when shopping starts inside an AI assistant.

What Imprint’s MCP server actually does

MCP is a way for AI applications to connect with information and tools held by other systems. Imprint’s server makes information associated with a cardholder’s account available to authorised AI agents, including rewards, virtual cards and transaction history. The announcement also describes tools that allow members to review and approve purchases initiated by an agent. Business Wire’s announcement and Scommerce’s coverage set out the product launch.

That combination is important. Rewards information can help an agent assess which card suits a purchase, while virtual cards can support a payment flow without exposing a cardholder’s usual card details to the agent. Transaction history may help the agent understand previous spending or identify relevant account activity.

The review and approval step is just as significant as the data access. Permission to read rewards is not automatically permission to spend. A member review step gives the cardholder a point to confirm or reject an agent’s proposed purchase before it proceeds.

There are still implementation details that businesses should verify rather than assume. The public announcement does not, by itself, explain every permission setting, spending limit or confirmation rule available through the MCP server. For merchants and platforms, the practical test is how a specific integration handles authentication, approval, transaction records and responsibility when an agent is involved.

A secure AI agent reads a rewards ledger and virtual payment tokens through a clear customer approval checkpoint

Rewards visibility can decide which card an agent picks

Imprint runs co-branded card programmes for brands including Kroger, Shell, Albertsons, Fanatics, H-E-B, Rakuten, Booking.com, Crate & Barrel and Fetch. The company says it has more than two million members and that its partner brands represent over $500 billion in annual US consumer spending.

The opportunity for these brands is not simply to put a logo on a card. It is to make the card relevant at the point an agent compares products and payment options. As Positive Sum CEO Patrick O’Shaughnessy put it: “AI will make every option easier to compare, which makes earning a customer’s preference more valuable than ever.”

If an agent can see one card’s relevant rewards and cannot access another’s, the first card has an advantage in the decision process. Brands relying on card programmes without an agent facing permission layer risk being invisible at that moment, even if their rewards are attractive to a human shopper.

That does not mean rewards alone will determine the outcome. Price, product availability, delivery, customer preferences and payment acceptance all matter. But rewards that cannot be read or applied by an agent may not enter the comparison at all.

Imprint CEO Daragh Murphy described the broader shift this way: “From AI chats to personal agents, every digital touchpoint is quickly becoming personalised.” For brands, that points to a change in the loyalty interface: an assistant may increasingly mediate the interaction that once took place on a retailer’s website, in an app or at the checkout.

An AI agent evaluates several payment choices while one loyalty reward signal illuminates the path to checkout

Funding capacity is part of the story

The $130 million financing is only one part of Imprint’s growth picture. The company also added $1.5 billion in incremental warehouse capacity and priced its second AAA rated asset backed securitisation. Investor demand for that securitisation was close to five times the offering, according to Scommerce and VCPost.

These funding mechanisms serve different purposes. Equity can support the business and product development. Warehouse capacity helps fund card balances, while securitisation packages eligible assets for investors. For a growing card platform, the ability to support lending and manage funding costs is relevant to how far a programme can scale.

Imprint also expects around $400 million in GAAP revenue by year end, up 150%. That is a company reported, unaudited expectation, not independently verified annual revenue. The company said cardholder spending across its portfolio more than doubled from 2025 to 2026. These figures indicate reported momentum, but they do not remove the need to assess credit performance, profitability and the durability of partner spending.

The new Imprint Access card broadens the strategy in another direction. Backed by a refundable deposit, it is intended for eligible people who are new to credit or rebuilding it, with a path to an unsecured product. That connects credit access to brand rewards, though the product’s long term performance will depend on customer outcomes and responsible credit management.

What Australian brands and platforms should do now

Australian retailers with loyalty programmes and co-branded cards should expect AI shopping agents to compare payment instruments, not just products. The same read and approve pattern is emerging through global network initiatives including Visa Intelligent Commerce and Mastercard Agent Pay. For more on how payment credentials may travel into agent led checkout, see Quantum Payments’ analysis of Shop Pay inside AI agents.

The strategic question is not simply whether an AI agent can make a payment. It is whether a brand’s value proposition is visible when the agent helps choose how that payment happens.

Frequently asked questions

What is Imprint’s MCP server?

It is a connection that allows AI agents to access specified information associated with Imprint card accounts, including rewards, virtual cards and transaction history. Members can review and approve purchases initiated by an agent.

Why does rewards visibility matter to co-branded cards?

An agent can only compare rewards it can access and interpret. If a card’s benefits are not available at the point of decision, the agent may favour a payment option whose rewards are visible.

What should Australian merchants do first?

Start by auditing whether loyalty rewards and payment rules are machine readable, then establish who is responsible for customer permission and approval. Test the full journey, from agent referral through payment and reconciliation.

The funding, product announcements and company growth figures were reported by Business Wire, VCPost, Scommerce and the AISeng Tech Payments series.

 
 
bottom of page