Who will gatekeep agentic commerce? From search and discovery to AI agents and wallets

Milko Filipov, Senior Manager for Financial Services at valantic, analyses the evolution of agentic commerce so far, highlighting the growing importance of digital wallets in an autonomous ecosystem.

The customer buying journey: from browsing to delegated commerce

Commerce is moving into a fundamentally different phase. For years, the model was straightforward: users searched, compared options, and completed purchases themselves. That is no longer the case. We’re seeing the early signs of delegated decision-making, where AI agents take on a more prominent role – interpreting user intent, narrowing down choices, and executing transactions. The customer journey is changing from purely user-driven to becoming agent-mediated.

For payments professionals, this shift is more than another interface evolution, as it changes where transactions are initiated and who controls them. Once an AI agent becomes the intermediary between the user and merchant, traditional checkout begins to lose its central role.  

In its place, a new critical layer emerges: the mechanism that translates intent into a trusted and authorised transaction. This is where wallets start to be positioned more strategically. In an agent-driven environment, they cease to be simply containers for credentials or payment methods and become the bridge between decision and execution.

Emerging gatekeepers in a machine-driven ecosystem

The question of who controls this new commerce flow cannot be answered by looking at a single player. Instead, it requires a more holistic view:

AI systems, AI agents, and the importance of clean merchant data for discovery

AI agents front this transformation, with large language models (LLMs) such as ChatGPT, Google Gemini, and Claude increasingly shaping how intent is interpreted, which options are considered, and how decisions are framed. In that sense, they do not simply replace search but abstract it, turning fragmented queries into structured outcomes. 

Search engines like Google remain critical, but their role is evolving from primary interfaces to infrastructures, providing data, rankings, and context that feed into AI systems. This has direct implications for merchants, in that visibility is no longer only about ranking in search results. Now, it depends on whether systems (not humans) can interpret and access your offering. 

In practice, this means structuring product data in a clean, consistent, and machine-readable manner, exposing it through APIs or standardised feeds, and aligning with emerging schemas used by AI systems. LLMs do not browse the web like users. Instead, they hinge on structured inputs, integrations, and retrieval mechanisms, which will, over time, likely favour merchants who optimise both for discovery and direct machine consumption.

The digital wallet: the bridge between intent and execution

Another layer gaining importance is the digital wallet. Emerging solutions such as Coinbase Agentic Wallets and Mastercard Agent Pay go past enabling payments and are beginning to determine whether an AI agent can make financial decisions. 

Payment providers such as Stripe, Adyen, and Checkout.com ensure that transactions can be executed, while commerce platforms like Shopify or Adobe (through Adobe Commerce) structure product data and enable integration. 

As these layers evolve, one observation becomes clear: control over execution is just as important as control over discovery – an area where wallets gain strategic relevance.

To understand their role, let’s first look at how agentic commerce works in practice: a user expresses intent in natural language, an AI agent interprets the intent, identifies the most relevant options, and reaches a decision. The process then shifts from decision-making to execution, where the wallet becomes essential, providing the components necessary to move from recommendation to action. It holds funds, stores credentials, enforces authorisation rules, and ultimately triggers the transaction. Without it, the agent can recommend, but it cannot act. 

At the same time, not all agentic transactions follow the same pattern. For example, in assisted scenarios, such as purchasing consumer goods, the user remains involved and makes the final decision, with the wallet acting as an execution layer following confirmation. 

In fully autonomous scenarios, however, the dynamic changes significantly. Agents act based on predefined rules or optimisation logic, for example, when purchasing datasets, allocating compute resources, or managing inventory. In these cases, the wallet evolves into a governance layer that defines what the agent is allowed to do, which brings a new level of complexity. The more autonomy is delegated to agents, the more responsibility shifts to the wallet to enforce limits, permissions, and conditions. 

For merchants, this raises new questions around trust, as accepting a transaction initiated by an AI agent is fundamentally different from accepting one initiated by a human. Concepts such as Know Your Agent are emerging, focusing on whether an agent is authorised, reliable, and operating within defined parameters. 

In this context, the wallet infrastructure itself becomes a trust proxy. Transactions routed through trusted providers (whether global payment networks like Mastercard or Visa or regulated wallet providers) are more credible than those from unknown systems, which positions wallets as potential trust anchors within agentic commerce.

Standards, protocols, and the battle for control in agentic commerce

As interactions across this ecosystem become more complex, the need for common standards and protocols becomes unavoidable. Agentic commerce depends on seamless communication between AI agents, wallets, merchants, and payment providers. Without interoperability, the model remains fragmented and difficult to scale. 

Early initiatives such as the Agentic Commerce Protocol (ACP), the Universal Commerce Protocol (UCP), or efforts like the x402 payments protocol show where the industry is headed. While the initiatives aim to define how intent is communicated, how authorisation is handled, and how transactions are executed across systems, the more important question, however, is who will shape these standards. 

Wallet providers are well-positioned to define identity and authorisation frameworks, AI providers to influence how intent is structured and decisions made, and payment providers to retain control over execution and settlement. The real leverage lies at the intersection of these layers. 

Simultaneously, the evolution of payment methods will be key in shaping how these standards develop. For consumer-driven, assisted transactions, existing payment rails, such as cards and digital wallets like PayPal, are likely to remain dominant due to trust, regulation, and familiarity. In contrast, autonomous and high-frequency machine-to-machine transactions may favour alternative approaches, such as stablecoins, programmable money, or even central bank digital currencies. 

The above suggests agentic commerce is unlikely to converge on a single universal model. Instead, different approaches will emerge depending on the use case: consumer vs enterprise, assisted vs autonomous, low-frequency vs high-frequency. Wallet providers face a clear strategic decision: whether to remain tied to existing payment schemes or to evolve into multi-rail platforms capable of supporting a broader range of agent-driven interactions.

What are the implications of agentic commerce for payments, wallets, and merchants?

For payments, wallets, and commerce leaders, agentic commerce holds significant implications: competitive advantage will not be defined within a single layer, but by their positioning within a broader stack that connects intent, trust, and execution. 

Wallets in particular are becoming central, evolving from passive instruments into active enablers of agent-driven commerce, capable of enforcing rules, managing risk, and establishing trust between machines and merchants. Being agent-ready will increasingly mean being able to support autonomous authorisation, integrate with AI systems, and operate across multiple payment rails. 

At the same time, it would be a mistake to assume that existing players will automatically dominate this space. Agentic commerce creates room for entirely new service categories, including machine-to-machine commerce platforms, agent identity layers, and transaction orchestration tools, innovations that are not just incremental, as they enable completely new forms of economic activity. 

In this type of environment, the ultimate gatekeepers may not be the current, evolved dominant players, but new layers that define how participants interact, from determining which wallets are trusted and which payment providers are integrated to which merchants are accessible. 

However, one thing is already clear: visibility alone will not suffice. In an environment where AI agents mediate transactions, success depends on being accessible to machines, trusted by the ecosystem, and executable within automated flows. 

About the author

Milko Filipov is a payments expert and Senior Manager in the Financial Services practice at valantic, and the driving force behind reMonetary. He specialises in payment systems, fintech, retail banking, and emerging technologies like CBDCs, embedded finance, and agentic commerce. With international experience at Deutsche Bank, Commerzbank, Visa, Stripe, and more, Milko helps organisations turn payments into strategic tools to drive revenue, innovation, and profitability.

 

About valantic

valantic is a fast-growing digital solutions, consulting, and software company that supports a broad range of international clients with their digital transformation journeys. Combining technological expertise with deep industry knowledge, valantic helps organisations turn strategy into execution, leveraging data, AI, and modern technologies to optimise processes, enhance customer experience, and drive measurable business outcomes.