AI agents, UCP and blockchain payments: the new e-commerce revolution has already begun

At a Christmas dinner with friends, conversations sometimes emerge that seem like pure technological speculation. People talk about the future, exchange convictions, imagine scenarios and, at that moment, everything appears to belong to the realm of ideas. But there are moments when reality begins to confirm, just a few weeks later, what had seemed like intuition. That is what happened when the conversation turned to the impact of AI on business and, in particular, to the disruption approaching e-commerce.

The starting point was simple, but deeply transformative: the online buying process, as we know it, should not remain tied to the classic funnel made up of Product Page, Cart and Checkout. For more than two decades, e-commerce has been designed around graphical interfaces built for human users. The customer searched, compared, clicked, added to cart, filled in details, selected a delivery method, validated payment and waited for confirmation. This sequence became so common that it almost feels natural. But it is not natural; it is simply the dominant architecture of a historical phase of the internet.

The rise of intelligent virtual assistants, AI-based conversational systems and agents capable of executing tasks on behalf of people and companies is shifting the centre of the digital experience. The user no longer has to navigate pages, filters, menus and buttons to reach a decision. Instead, they can express an intention: «I want to buy a gift for a marketing manager, up to 80 euros, with delivery by Friday». From that request, an AI agent may search for alternatives, compare criteria, validate availability, check delivery conditions, choose the most suitable option and initiate the purchase.

What changes, then, is much more than the interface. The infrastructure of the purchase itself changes. The journey no longer depends only on a visual experience and begins to rely on protocols, permissions, structured data, APIs, trust mechanisms and programmable payment systems. What previously happened on the surface, through pages and forms, starts moving into backend processes, where autonomous agents communicate with catalogues, platforms, financial services, logistics networks and verification systems.

From the visual funnel to the invisible funnel

E-commerce has always been described through funnels. First, discovery. Then, consideration. Next, intention. Finally, purchase. This logic remains useful for marketing teams, but it is becoming insufficient to explain what comes next. In an environment mediated by AI agents, the funnel does not disappear, but it becomes less visible. Product comparison, price evaluation, review analysis, store trust assessment and condition validation may all happen without the user visiting a single product page.

This scenario represents a radical change for brands, retailers and e-commerce platforms. Until now, much of digital optimisation has focused on improving visual elements: product photography, page hierarchy, call-to-action buttons, site speed, simple forms, cart recovery systems and clarity at checkout. All of this will continue to have value, because humans are not going to disappear from the buying experience. But it will no longer be enough.

If a growing share of decision-making becomes mediated by agents, stores will have to be understandable to machines, not just persuasive to people. The catalogue will need to be better structured. Product data will need to be accurate. Commercial policies will have to be readable by automated systems. Availability, price, payment methods, delivery and returns will have to be exposed in a clear, secure and interoperable way.

This is where the conversation about UCP, presented by Shopify in collaboration with Google, becomes especially relevant. The idea of a protocol designed for agent-led purchases points to a future in which consumer intent can be interpreted by intelligent systems and converted into concrete actions within commercial environments. Purchasing stops being just a sequence of clicks and becomes a transaction negotiated between systems.

UCP: e-commerce prepared for agents

UCP emerges as an answer to an essential question: how can an AI agent buy on behalf of a user in a secure, verifiable way that is compatible with commercial platforms? The question may seem technical, but it has deep strategic implications. If agents become intermediaries between consumers and stores, a common language will be needed so that everyone can understand requests, products, restrictions, permissions and payments.

In a traditional model, the website is the main point of contact. In an agent-oriented model, the site becomes just one of the entry points. A store may receive human traffic through search engines, paid campaigns, social media and email, but it may also receive requests from personal assistants, AI platforms, corporate systems or vertical applications. The product is no longer found only by a person; it is interpreted by a machine acting according to an intention.

For those working in e-commerce, this change requires a new mindset. An online store should not be seen only as a digital storefront, but as a transactional system. The quality of the visual experience remains important, but data quality, integration robustness, clarity of commercial rules and API readiness gain increasing weight.

The consequence is clear: the future of e-commerce will be more technical, more automated and more dependent on trust. A brand that sells online will have to prove, to both humans and agents, that its product is relevant, its data is reliable, its price is up to date, delivery is possible and the transaction is secure. Marketing starts to live in a hybrid layer: communication for people and legibility for systems.

The role of programmable payments

At the dinner mentioned above, the conversation gained an even more interesting dimension because one of the people involved was connected to the financial sector. The question was no longer only «how does an agent choose a product?» but became «how does an agent pay?». This is one of the central questions of the next phase of the digital economy. An AI agent can search, compare and recommend, but the purchase only becomes real when an appropriate payment infrastructure exists.

Traditional payment methods were created for humans, accounts, cards, manual authorisations, banking interfaces and authentication processes. They are extremely relevant systems and will remain so, but they were not designed from the outset for autonomous agents that can make micropayments, pay per request, settle services in fractions of a second or interact with multiple APIs without monthly subscriptions.

This is where stablecoins, blockchain and machine-oriented payment protocols enter the picture. The news about Solana Foundation and Google Cloud, with the launch of Pay.sh, is especially significant because it connects several pieces of the puzzle: AI agents, per-request payments, APIs, cloud services, open protocols and settlement through stable digital assets.

According to the reference text, Pay.sh allows AI agents to autonomously pay for Google Cloud services and other APIs with stablecoins on the Solana network. Instead of a monthly subscription or a traditional account, an agent can pay fractions of a cent per API call. This logic is much closer to the way autonomous systems may operate: granular consumption, immediate payment, traceability and direct integration between usage and settlement.

Why blockchain makes sense in this context

The association between blockchain and e-commerce has not always been well understood. For years, many proposals seemed to look for problems in order to justify a technology. However, when we talk about autonomous AI agents, micropayments and machine-to-machine transactions, blockchain starts to fit more naturally.

There are several reasons for this. First, because agents need a programmable way to transfer value. Second, because payment execution should be able to happen without always depending on manual processes. Third, because transparency and verifiability are important when decisions are delegated to automated systems. Fourth, because very low-value payments, per request or per use, can be difficult to support on traditional financial infrastructures with high fixed costs.

Stablecoins add an important layer to this scenario because they reduce the volatility associated with many cryptoassets. For operational payments, stability is essential. A company does not want the cost of an API to vary drastically because of an asset’s price fluctuation. An AI agent also needs predictability to manage budgets, limits and permissions. In this sense, stablecoins and autonomous agents can form a practical combination for certain use cases.

This does not mean that all future payments will be made on blockchain, nor that traditional financial systems will be replaced overnight. It means, rather, that there are new transaction models for which existing infrastructures may not be sufficient. And when new models emerge, new technological layers also emerge.

x402, MPP and the API call economy

The reference text mentions the x402 protocol, presented as an open AI payments standard, initially incubated by Coinbase and later under the stewardship of the Linux Foundation. It also refers to the Machine Payments Protocol, developed by Tempo and Stripe. Even without entering into an exhaustive technical analysis, the direction is clear: the industry is building mechanisms so that machines can autonomously buy digital services.

This change may seem distant from online retail, but it is not. Today, an online store already depends on a wide range of services: hosting, internal search, recommendations, fraud assessment, logistics, invoicing, email marketing, data analysis, personalisation, advertising, translation, description generation, chat support and much more. If each of these layers becomes dynamically consumed by agents, pay-per-use becomes a natural model.

Imagine a commercial agent managing an online store overnight. That agent checks stock data, adjusts campaigns, requests demand forecasts, generates description variations, validates competitor prices, purchases computing capacity to process reports, pays for calls to language models and activates external services only when needed. Each action may generate a minimal cost, settled automatically within previously defined limits.

The API economy therefore stops being limited to subscriptions or rigid contracts. It starts working as a granular market, where agents pay for each unit of value consumed. For digital businesses, this logic can be powerful, but it also requires governance. Permissions, spending caps, auditing, approval criteria, reversal mechanisms and security rules will need to be defined.

The purchase stops being a page and becomes an intention

For a long time, selling online meant attracting users to a page. Hence the importance of traffic, conversion rate, user experience and checkout optimisation. These elements do not disappear, but they begin to coexist with a new reality: selling online will also mean making products eligible for agents.

When a consumer asks an AI assistant to find the best product for a specific need, the brand no longer competes only for visual attention. It competes for semantic relevance, data quality, reputation, availability, price, trust and technical compatibility. The agent may not be influenced by a beautiful photograph or an emotional sentence, but it may value structured attributes, consistent reviews, clear policies and authority signals.

This change will have a direct impact on SEO. Search engine optimisation can no longer be thought of only in terms of keywords, pages and rankings. It will have to include structured data, entities, intent, topical authority, informational quality and preparation for generative answer environments. The question stops being only «how do I appear on Google?» and also becomes «how am I understood, selected and recommended by agents?».

For e-commerce managers, this is a strategic shift. Products will have to be described precisely. Categories will need to make sense. Variations cannot be confusing. Information about sizes, compatibility, materials, certifications, warranties and availability should be treated as a commercial asset. Content stops being decoration and becomes infrastructure.

From human persuasion to algorithmic trust

Digital marketing has always sought to influence human decisions. Campaigns, messages, creatives, landing pages and sales arguments were created for that purpose. In a world mediated by agents, persuasion does not disappear, but it gains another layer: algorithmic trust. An agent needs reasons to recommend a brand. Those reasons may include price, quality, history, reviews, delivery speed, reputation, return policies, sustainability, compatibility with user preferences or data reliability.

This means that branding will remain relevant, but it will be interpreted in new ways. A strong brand can influence consumer preference, but it also needs to provide objective signals that an agent can process. Authority will not be only narrative; it will also be technical. A store with inconsistent data, slow pages, ambiguous policies or fragile integrations may lose opportunities even if it has good products.

Trust starts to be built across multiple layers. There is consumer trust in the brand. There is agent trust in the data. There is platform trust in the protocols. There is financial system trust in the authorisation. There is merchant trust in the agent’s scope of action. Each of these layers will have to be planned, configured and monitored.

In this context, the role of digital agencies also changes. It is no longer enough to create visually appealing campaigns or good-looking stores. It is necessary to design commercial ecosystems capable of communicating with people, search engines, AI platforms, payment systems and data infrastructures. Creativity remains essential, but it needs to live alongside technical architecture.

The impact on Shopify and online selling platforms

Shopify has a particularly interesting position in this scenario because it brings together millions of merchants, a robust app ecosystem, payment integrations and a strong development culture oriented towards commerce. When a platform of this scale participates in defining protocols for agent-led purchases, the signal to the market is clear: e-commerce is preparing for a new consumption interface.

The future should not be only about more beautiful stores, but about more interoperable stores. The ability to expose products, rules, prices and availability to external agents may become a competitive advantage. Just as a store today needs to be optimised for mobile, tomorrow it may need to be prepared for AI-initiated purchases.

This evolution also raises important questions. Who controls the final decision? How can we ensure that the agent respects user preferences? What responsibility exists if a purchase is executed incorrectly? How are returns validated? How is fraud prevented? How can agents be prevented from making unauthorised purchases or exceeding budgets? These questions show that the revolution is not only technological; it is also legal, financial and operational.

Despite the doubts, the direction seems inevitable. Whenever a technology reduces friction, increases convenience and creates new economic models, the market tends to explore it. The traditional checkout will not disappear immediately, but it will stop being the only legitimate way to complete a purchase. Consumers may buy within a conversation, from a personal assistant, through a voice device, via a business agent or inside an application that never opens the store in the classic sense.

What changes for businesses?

For companies, the first change is mental. The online store stops being seen as the final destination and starts being seen as a node in a network. Products can be discovered, evaluated and purchased outside the visual environment controlled by the brand. This requires a more open, more structured and more data-oriented strategy.

The second change is operational. Businesses will have to review integrations, catalogue quality, APIs, payment systems, automations, stock rules, attribution models and security. An information error that today only causes a poor user experience may, tomorrow, lead an agent to automatically exclude a brand from a recommendation.

The third change is competitive. Brands that prepare their digital assets early for AI agents may gain an advantage. Not because all consumers will immediately start buying through assistants, but because the transition will be gradual. First, specific cases will emerge: repetitive purchases, replenishment of consumables, bookings, digital services, API acquisition, B2B products, purchases with clear criteria. Then, the logic will expand into more complex experiences.

The fourth change is financial. If agents can pay for services, products or data with stablecoins or other programmable means, companies will have to understand which infrastructures they accept, what risks exist, how reconciliation is done, which tax obligations apply and how these flows are integrated into management systems. The technology may seem simple at the experience layer, but it requires maturity behind the scenes.

The new role of digital marketing

Digital marketing will not be replaced by AI agents, but it will be profoundly reconfigured. Traffic acquisition will continue to exist, but part of demand may move towards conversational systems. Advertising will continue to have value, but it will need to adapt to contexts where decisions are filtered by intelligent intermediaries. Content will remain essential, but it will have to be useful for humans and interpretable by machines.

Digital marketing campaigns will need to consider new questions: what data feeds the agents? Which attributes differentiate the product? What trust evidence is available? What content responds to complex intentions? Which integrations allow a purchase to be executed without friction? What limits should be defined for commercial automations?

Metrics will also need to be reconsidered. Page conversion rate will remain relevant, but it may not capture purchases initiated outside the site. Click-based attribution may become less clear. The value of content may appear in recommendations made by agents, not only in organic visits. Brands will have to develop new ways of measuring presence, influence and preference in AI-mediated environments.

This transition requires the unification of teams that, in many companies, still work separately: marketing, technology, data, sales, customer service, finance and legal. Agent-led purchasing is not just a feature. It is a business model that crosses communication, infrastructure and trust.

Why the Google, Solana and Shopify vision makes sense

The connection between Google Cloud, Solana, protocols such as x402, initiatives such as UCP and the Shopify ecosystem points towards the same horizon: the creation of an internet where agents can act, buy, pay and consume services autonomously, within limits defined by humans. This vision may not yet be the market standard, but it has all the ingredients to influence the next decade of digital commerce.

Google has the cloud, AI and distribution infrastructure. Shopify has the commercial ecosystem. Solana offers a network geared towards fast transactions and low costs. Coinbase, Stripe, the Linux Foundation and other players help create standards and trust mechanisms. None of these pieces, in isolation, solves the problem. Together, they begin to outline a new economic layer for the internet.

The essential point is that AI agents need more than intelligence. They need access, permissions, identity, budget, payment methods and the ability to execute actions. An assistant that recommends but cannot act is only a consultative tool. An agent that understands an intention, validates options, pays for resources and completes a transaction enters another territory: economic automation.

That is why the conviction shared at that Christmas dinner becomes relevant. The idea that blockchain would play an important role was not based on abstract enthusiasm for cryptoassets, but on a functional reading of the problem. If agents need to pay each other, per request, with traceability and low cost, then a programmable settlement infrastructure becomes a logical hypothesis.

The end of checkout as the centre of the experience?

It would not be correct to say that checkout is going to die. Most likely, it will stop being the only centre of the buying experience. It will continue to exist in many contexts, especially when consumers want to manually control every detail. However, in recurring purchases, standardised products, digital services, B2B, automatic replenishment and decisions based on objective criteria, the visual checkout may become secondary.

The new centre of the experience will be intention. The user defines what they want, their limits and preferences. The agent interprets, searches, compares, negotiates and executes. The interface may be a conversation, a voice instruction, an automated rule or a business integration. The act of buying becomes less dependent on a sequence of pages and more dependent on delegated trust.

For brands, this creates both an opportunity and a threat. The opportunity lies in reducing friction and reaching consumers in new contexts. The threat lies in losing control over the visual narrative of the purchase. If an agent presents three summarised options, the brand needs to ensure that its most important attributes are recognised and valued. The product must be good, but it must also be understandable to automated systems.

Preparing today the online store of the future

Companies do not need to wait for mass adoption of UCP, x402 or other protocols to start preparing. There are practical measures that already make sense: better organising product data, reviewing taxonomies, improving descriptions, structuring technical information, ensuring consistency across channels, optimising speed, strengthening security, creating solid integrations and documenting commercial rules.

It is also important to invest in content that responds to real intentions. AI agents tend to value context, clarity and authority. A store that explains its products well, compares options, answers questions, publishes useful guides and maintains coherent information will be better positioned to be interpreted by intelligent systems. Content stops being used only to attract traffic; it starts feeding decisions.

Another essential step is reviewing the technological architecture. Platforms such as Shopify offer a solid foundation for growth, but competitive advantage depends on how they are configured, integrated and adapted to the business model. The next phase of e-commerce will be less tolerant of disorganised catalogues, incomplete data and improvised integrations.

Finally, companies should start discussing decision delegation internally. Which purchases can be automated? What limits should exist? What data can be shared with agents? Which payments can be executed without human intervention? What auditing is necessary? These questions will be as important as current decisions about campaigns, channels and budgets.

The e-commerce revolution has already begun

The history of digital commerce is made of moments when a new layer changes everything that seemed stable. Search changed the way we discover products. Social media changed the way brands create desire. Mobile changed the way we buy from anywhere. Marketplaces changed distribution. Now, AI and programmable payments may change the execution of the purchase itself.

What is most interesting is that this revolution is not limited to the end consumer. It will also transform B2B, digital services, data acquisition, API consumption, infrastructure management and business automation. The same principle applies in multiple contexts: agents with defined goals, access to services, payment capacity and rules of action.

The vision is still under construction. There will be competing standards, failed experiments, regulatory challenges, security issues and cultural resistance. But the direction is too strong to ignore. When technology giants, commerce platforms, open-source foundations, financial companies and blockchain networks move to solve the same problem, it is a sign that the market is preparing a new infrastructure.

For those who sell online, the question is no longer only when this reality will arrive. The right question is: which part of the business is prepared to be discovered, understood, recommended and transacted by AI agents? Companies that start structuring data, integrations, content, payments and trust now will be better positioned to compete when this new layer becomes common.

The conviction shared at that dinner was, after all, less futuristic than it seemed. Online shopping is moving from navigation to intention, from interface to protocol, from click to automation, from cart to intelligent execution. E-commerce will not simply receive AI as another feature. It will be redesigned by it.

BYDAS follows this evolution with experience in digital strategy, web development, Shopify and performance. If your brand wants to prepare a more robust, integrated online store ready for the next generation of digital purchases, discover our Shopify solutions.