AI platforms have already become a key entry point for shopping. Our latest research shows that 59% of US consumers are now using AI for product research, up from 36% in March 2025. But that’s just the start of a longer adoption process. AI agents that can act independently (with human approvals and controls) will soon give consumers the ability to automate the whole shopping journey, including payment execution. By 2030, agentic commerce may constitute as much as a fifth of e-commerce in the US, according to Forrester.

Amid this profound change, financial services incumbents can’t take their role in e-commerce for granted. AI platforms are already becoming the primary online interface for many shoppers and are increasingly looking to exert end-to-end influence over the shopping journey, from product discovery to checkout. What’s more, they appear to have been building brand equity in e-commerce with remarkable speed.

When we asked US consumers in March 2025 how comfortable they would be using AI for purchases, they seemed hesitant to rely on the likes of ChatGPT, Claude, and Gemini. Instead, consumers signaled they would be more comfortable buying with AI tools if the tools were associated with familiar retail and payment brands. Now, having experimented more with AI tools and, increasingly, AI agents, US consumers seem more willing to shop with the newcomers.

Although shopper attitudes to AI are still evolving, this shift in consumer trust highlights a strategic reality: The e-commerce profit pools of card issuers, payment service providers (PSPs), and banks are at risk of significant disruption, and the advantages of incumbency alone won’t fully insulate those players from being relegated to the margins of agentic commerce. The danger is fourfold:

1. AI agents might default to lower-margin payment methods, such as debit cards and account-to-account.
2. When agents begin to optimize choice of payment method in addition to handling search and discovery, AI platforms might demand a greater share of the payments profit pool to reflect their expanded role in the shopping journey.
3. Agents could compress the economic model of credit cards by helping consumers optimize which cards to use to maximize rewards and redemptions.
4. By reducing the friction of using card rewards and benefits, AI agents could put pressure on card product design that historically has had built-in barriers to consumer value maximization.

These are just the threats that are most apparent today in agentic commerce’s embryonic state. More might emerge. The good news is that, with the right moves, card issuers and banks can start to integrate their unique value proposition into this new shopping experience, maintaining both their visibility and their direct relationship with consumers.

How to remain top of wallet in agentic commerce

Agentic commerce presents a familiar challenge to financial services incumbents: remaining “top of wallet” in e-commerce transactions. For instance, when payments first went online in the 1990s, card issuers and banks had to jostle for prominence at the point when a shopper inputted their card number at checkout. Similarly, when digital wallets were created, the challenge for those incumbents was to be chosen as the default card in the wallet.

With their centrality to e-commerce once more at risk, card issuers and banks can take a little consolation from our new research, which shows that credit and debit cards are, for now, the preferred payment methods for the AI shopping journey (see Figure 1).

Figure 1

To capitalize on this advantage as agentic commerce matures, card issuers and banks need to keep up with what people want most when using this new channel, both in terms of their direct preferences for today’s AI-assisted purchases and the delegated mandate that they are likely to give AI agents for fully automated transactions in the future. For instance, our research suggests that there’s an opportunity to preserve profitability through the provision of agentic purchase protection, transaction thresholds, and other security measures (see Figure 2).

Figure 2

To remain top of wallet in the agentic commerce era, players within the card ecosystem will also need to proactively focus on three imperatives: ensuring that their payment methods are fully visible and enabled, adapting their unique value proposition and products, and defining their posture toward big AI platforms.

Ensuring payment methods are fully visible and enabled. AI agents won’t be able to optimize payment method if they don’t see the rewards and benefits. Consequently, ensuring that terms, benefits, and other key information are readable to AI (in addition to being able to accept AI-driven transactions) is essential for financial institutions. Crucially, a card’s rewards and offers need to show up early in the shopping journey, when an agent is searching for and comparing products, rather than just at checkout, when it’s often too late. Big banks can go further, integrating directly with leading AI agent platforms to influence preferred-card placement and showcase differentiated card features, such as anti-fraud measures.

Adapting unique value proposition and products. Visibility won’t always be enough. Incumbents will sometimes need to redesign a rewards program or even an entire card program to ensure that their unique value proposition is projected well within agentic commerce. For instance, a card issuer might need to move away from offering fixed rewards to a more dynamic and contextual approach that is better able to influence routing decisions by agents. Most banks and other card issuers will have mountains of proprietary customer data that can contribute to a great agentic experience—by decoding intent-to-buy signals or by making it easier to finance a purchase, for instance.

Defining posture toward big AI platforms. For many financial institutions, constraints of scale and proprietary assets will limit their options to create differentiated agentic commerce experiences for consumers, above and beyond what the AI platforms can create. However, some incumbents will have the assets and resources to be competitive as well as cooperative. For instance, if a large financial institution has a popular service affiliated to its cards—a travel site, say—it might choose to roll out AI tools on that site in a way that is distinctive and maximizes the site’s standalone value proposition, rather than pursuing maximum integration with the likes of ChatGPT and Gemini. Preserving that unique customer experience might mean holding back some features from other channels.

Above all, it will be vital for card issuers, PSPs, and banks to take a proactive rather than reactive stance as they develop their strategic response to the coming disruption. Consider a card issuer that, rather than waiting for disruption to arrive, proactively defined its agentic commerce strategy in early 2026. It is now working with its merchant brand partners to bring integrated agentic AI checkout to market in a way that emphasizes the benefits and value proposition of its cobranded credit cards.

Getting the tech foundations right for agentic commerce

It’s still early days with agentic commerce. Most usage of generative AI tools in shopping is in the research and comparison phase. The next step for these shoppers isn’t to delegate purchase decisions to AI agents but first to execute payment within an AI chat interface on a user-authenticated basis.

However, a more autonomous future is likely to follow quickly, first with agents making purchases on behalf of shoppers, then on an agent-to-agent basis that takes automation one step further. Amid heavy investment by AI platforms in the protocol infrastructure needed by agentic commerce, innovative issuers and financial institutions are already working to enable shoppers to execute payments within AI-based apps, without having to be diverted back to a regular e-commerce checkout.

In addition to making strategic choices about how to compete in agentic commerce, banks and other card issuers must also develop the technological infrastructure and capabilities to realize their vision. That means deepening expertise in areas such as authentication, identity linking, and secure management of an increasingly complex set of commerce protocols (such as Google’s UCP and OpenAI’s ACP). Tomorrow’s winners will almost certainly combine a clear vision for agentic commerce with the tech savvy to build compelling new customer experiences and leverage valuable data.