The following interview explores the rapid evolution of agentic commerce, a paradigm shift where artificial intelligence moves beyond simple suggestions to autonomous execution. Priya Jaiswal, a leading expert in banking and international finance, joins us to discuss how this technology is redefining the relationship between consumers, brands, and the underlying financial infrastructure. We delve into the critical role of open protocols, the necessity of programmable currencies for machine-to-machine transactions, and the complex trust frameworks required to let algorithms manage our wallets.
How do you distinguish between a standard AI shopping assistant and a system with true agency? What specific multi-step workflows, such as identity authentication or post-purchase service, must an agent handle to move beyond mere recommendations into full transaction execution?
The fundamental difference lies in the transition from advice to action. A standard shopping assistant is essentially a sophisticated search engine that points you toward a product, but a system with true agency possesses the authority to close the loop on a transaction. To achieve this, an agent must navigate a complex, multi-step workflow that includes comparing multiple options against a user’s specific constraints, such as a budget of exactly $150 or a preference for blue stripes on shoes. It doesn’t stop at the “buy” button; the agent must be able to authenticate the user’s identity, arrange the payment through secure APIs, and even handle the “unseen” logistics of post-purchase service like tracking a delivery or initiating a return. We are moving toward a model where the consumer specifies an intent, and the agent operates within explicit guardrails to fulfill that intent entirely, rather than just providing a list of links for the human to click.
As consumers shift away from manual browsing, how should businesses restructure their product data for Answer Engine Optimization (AEO)? What specific technical steps ensure that an autonomous agent can accurately evaluate a brand’s inventory and pricing against highly specific consumer constraints?
For years, businesses focused on traditional search engine optimization to attract human eyeballs, but the rise of agentic commerce requires a pivot toward Answer Engine Optimization (AEO) and LLM Engine Optimization (LEO). This means structuring product data into machine-readable formats that allow AI agents to instantly ingest and evaluate inventory, pricing, and business policies in real time. Businesses must ensure their data is not just “online” but is accessible via open protocols that provide shared context, allowing an agent to see real-time stock levels or specific delivery windows without human intervention. If a brand’s data is siloed or unstructured, an agent will simply bypass it because it cannot verify if the product meets the highly specific constraints—like a specific material or a guaranteed Tuesday delivery—set by the consumer. It’s a shift from aesthetic web design to high-fidelity data integrity that serves the logic of a machine.
Current digital commerce often relies on fragmented, custom integrations that slow down innovation. How do open standards like the Universal Commerce Protocol (UCP) or Agent Payments Protocol (AP2) create a “common language” for transactions, and what does this mean for interoperability between competing platforms?
The current state of digital commerce is a “patchwork of custom plugs for custom sockets,” which is inherently slow and expensive to maintain. Open standards like the Universal Commerce Protocol (UCP) act as the connective tissue, creating a platform-agnostic language that allows consumer surfaces, payment providers, and business systems to talk to each other seamlessly. For instance, AP2 provides the “secure handshake” necessary for payments, ensuring that when an agent initiates a transaction, there is a foundation of trust and accountability across the entire ecosystem. This interoperability means that a shopping journey started on a search engine can be completed within a bank’s app or a retailer’s private platform without losing the context of the user’s loyalty rewards or shipping preferences. By removing the friction of fragmented integrations, we enable a world where agents can operate across the retail value chain regardless of which tech giant’s ecosystem the user started in.
Trust in AI for financial decision-making remains quite low among the general population. How can “Know Your Agent” (KYA) frameworks and cryptographic signing bridge this gap, and what specific human-in-the-loop safeguards are necessary to prevent unauthorized or harmful high-risk actions?
With only about 10% of the U.S. population currently trusting AI with their financial decisions, bridging the trust gap is perhaps our biggest hurdle. We address this by implementing “Know Your Agent” (KYA) frameworks and cryptographic signing, which provide a verifiable digital identity for every agent, ensuring it is a legitimate actor with specific, limited authority. To prevent unauthorized actions, we must use graded permissions and spending limits, alongside “human-in-the-loop” checks where the system pauses for explicit consent before executing high-risk or high-value transactions. This ensures that while the agent does the heavy lifting, the human remains the ultimate “merchant of record” or decision-maker. Additionally, we are seeing the development of agents that can explain their recommendations and surface trade-offs, helping users understand why a specific option was chosen and ensuring that the AI’s logic is transparent rather than a black box.
Programmable currencies and stablecoins are frequently positioned as the essential settlement layer for machine-to-machine interactions. Why is traditional banking infrastructure often ill-suited for the sub-millisecond state access these agents require, and how does blockchain provide a more transparent, immutable alternative for automated commerce?
Traditional banking databases and legacy infrastructures were simply not designed for the sub-millisecond state access and real-time coordination that a world of billions of autonomous agents requires. When machines transact with machines, they need a settlement layer that is 24/7, programmable, and capable of near-instant finality, which is where stablecoins and blockchain technology become critical. Blockchain serves as a fully transparent and immutable ledger, allowing agents to verify funds and settle transactions without the multi-day delays of traditional clearinghouses. By using programmable currency, we can embed the rules of a transaction directly into the payment itself—for example, the funds only release once the agent confirms the shipping label has been generated. This turns the payment layer into a proactive part of the commerce workflow rather than a reactive, slow-moving administrative hurdle.
Regulations like the EU AI Act and eIDAS 2.0 create a complex patchwork of requirements for global commerce. How can developers embed policy-aware routing into agent workflows to ensure compliance across different jurisdictions without sacrificing the “invisible” nature of a seamless checkout experience?
Navigating global regulations like the EU AI Act or data residency laws requires a sophisticated technical approach known as policy-aware routing. This involves embedding compliance logic directly into the agent’s decision-making engine, so that its actions—such as how it handles identity checks or where it stores transaction data—automatically adapt based on the user’s jurisdiction. For example, an agent could use local execution for sensitive data to satisfy privacy laws while using network-level controls for fraud and risk detection. The goal is to make these regulatory “checks” part of the background processing, maintaining the “invisible” nature of the checkout while ensuring the transaction is legally sound. By automating these compliance hurdles, developers can allow agents to scale globally without requiring the end user to understand the nuances of international trade or data privacy laws.
In a future where algorithms negotiate and settle transactions before a human ever sees a receipt, how does the retail value chain fundamentally change? What happens to traditional brand loyalty when a “supervisor agent” is making purchasing decisions based purely on logic, preferences, and constraints?
The retail value chain is moving toward a model of “intent-based purchasing,” where the gap between interest and ownership is virtually eliminated. In this future, we may see a transition from single-purpose assistants to “teams” of specialized agents—one for retrieval, one for payments, and one for risk—all coordinated by a “supervisor agent” that manages the entire process. Traditional brand loyalty faces a significant challenge here, as a supervisor agent makes decisions based on cold logic, comparing real-time pricing, availability, and alignment with the user’s specific constraints rather than being swayed by flashy advertisements. However, this also opens the door for “Brand Agents” that live on a retailer’s site, trained on their specific catalogue and business rules to provide an authentic, on-brand conversational experience. The focus for brands will shift from capturing human attention to providing the best verifiable value that an agent can logically recommend.
What is your forecast for agentic commerce?
I forecast that by 2026, agentic commerce will have transitioned from experimental pilots—like the current trials being run by major players like Visa, Mastercard, and Santander—into a foundational element of the global economy. We will see the “invisible checkout” become the standard, where the labor of shopping is replaced by a continuous, background service that manages our subscriptions, handles our returns, and proactively finds deals before we even realize we need them. As programmable currencies and 24/7 payment infrastructures like stablecoins become more integrated with agentic platforms, we will witness a world where commercial transactions are routinely negotiated and settled between algorithms. Ultimately, as time becomes an increasingly precious commodity, consumers will view the ability to delegate their purchasing power to a trusted, logic-driven agent as an essential utility rather than a luxury.
