Online shopping is beginning to move beyond search boxes, product filters and conventional checkout pages. The next shift is agentic commerce: a model in which an AI agent can understand what a customer wants, research products, compare options, interact with a merchant, build a cart and, when appropriately authorised, help complete the transaction.
For an Indian consumer, this could eventually be as simple as saying, “Find running shoes under ₹4,000, make sure they arrive before Saturday, compare the return policies and buy the best option.” Instead of returning a list of links, an AI agent could potentially complete much of that workflow on the customer’s behalf.
India is particularly important to the development of agentic commerce because it combines a huge online-shopping population with UPI, quick commerce, sophisticated payment companies and rapidly improving AI infrastructure. In August 2026 alone, UPI processed approximately 24.51 billion transactions worth ₹29.82 lakh crore across 752 participating banks.
Agentic commerce in India is no longer purely theoretical. Swiggy has introduced AI-native ordering integrations, Mastercard has demonstrated authenticated agentic transactions in India, Pine Labs says it has deployed an autonomous UPI payment protocol, and NPCI is developing a registry intended to verify AI agents making UPI transactions. However, businesses should not interpret these developments as meaning fully autonomous AI shopping is already available across every Indian merchant or every UPI transaction. The ecosystem is still developing.
What Is Agentic Commerce?
Agentic commerce is digital commerce in which an AI agent can perform one or more commercial actions on behalf of a consumer or business, rather than only providing information or recommendations. These actions may include discovering products, comparing alternatives, checking availability, creating a cart, applying user-defined conditions and initiating or completing an authorised payment.
The difference between ordinary AI shopping assistance and agentic commerce is therefore action. A conventional AI assistant might tell you which three smartphones appear suitable for your budget. An agentic system can potentially move from understanding the request to interacting with commerce systems and executing the next permitted steps.
Visa describes agentic commerce as experiences in which AI agents help consumers or businesses discover products, make decisions and complete parts of the purchasing journey, ranging from product comparison and checkout initiation to transactions completed with user permission.
Agentic Commerce in India: Quick Answer
As of September 10, 2026, agentic commerce is operating in India through a mixture of live integrations, controlled implementations, payment infrastructure and pilots. It has moved beyond demonstrations of AI product recommendations, but a universal system where any AI agent can automatically purchase from any Indian merchant using UPI does not yet exist.
Swiggy introduced Model Context Protocol integrations across Food, Instamart and Dineout in January 2026, allowing users to interact with its commerce services through AI tools such as ChatGPT, Claude and Gemini. Instamart’s implementation launched with access to more than 40,000 products, including the ability to search products, build carts, apply coupons and track deliveries through conversational interfaces.
India’s payments ecosystem is developing simultaneously. Mastercard completed an authenticated agentic-commerce transaction in India in February 2026 using cards issued by Axis Bank and RBL Bank, while Pine Labs announced its P3P protocol in June and says it enables AI agents to complete eligible UPI transactions under pre-defined permissions. NPCI is now developing an AI-agent registry as part of a planned Unified Agentic Protocol for UPI.
How Agentic Commerce Is Different From Normal E-commerce
| Commerce Model | Customer Experience | Role of AI | Transaction |
|---|---|---|---|
| Traditional e-commerce | User searches, compares and clicks manually | Recommendations may assist | User completes checkout |
| Conversational commerce | User shops through chat or voice | Answers questions and guides discovery | Often still user-controlled |
| AI-assisted commerce | User provides requirements | Researches and recommends suitable products | User normally chooses and pays |
| Agentic commerce | User provides a goal and constraints | Can research, use commerce tools and perform permitted actions | Human-approved or delegated |
| Autonomous agentic commerce | User gives instructions in advance | Agent may act later within defined limits | Uses pre-authorised controls |
A chatbot therefore should not automatically be described as an agentic-commerce system. The important question is whether the software can safely interact with real commerce infrastructure and perform actions such as retrieving current inventory, creating orders or initiating authorised payments.
Why India Is an Important Market for Agentic Commerce
Agentic commerce requires more than advanced AI models. It needs digital consumers, structured merchant information, reliable payment infrastructure, fulfilment systems and enough transaction volume to justify automation. India already has many of those foundations at significant scale.
India’s online shopper base reached approximately 290–300 million people in 2025, according to IBEF. Tier-2 and smaller cities account for roughly 65% of new online shoppers, which means the opportunity is not restricted to major metropolitan markets. India’s online retail market reached approximately $80 billion in FY26, while quick commerce alone had already become a $7–8 billion market in FY25.
A September 2026 IBEF update projects India’s broader e-commerce market could grow from approximately $125 billion in 2024 to $345 billion by 2030, equivalent to an estimated 18.4% compound annual growth rate under that report’s market definition. Quick commerce could reach $65–70 billion by 2030. Forecasts from different organisations use different definitions of e-commerce and e-retail, so these figures should be treated as market projections rather than guaranteed outcomes.
UPI adds another major advantage. NPCI recorded 24.51 billion UPI transactions in August 2026, compared with 22.35 billion in April. This means India does not need to invent digital payments specifically for AI agents; much of the opportunity involves developing secure identity, permission and delegation mechanisms on top of payment infrastructure consumers already use.
India’s Agentic Commerce Ecosystem in 2026
| Company / Infrastructure | Agentic Commerce Development | Status |
|---|---|---|
| Swiggy | Food, Instamart and Dineout commerce through AI assistants using MCP | Live integration |
| Mastercard | Authenticated agentic transactions using tokenised card infrastructure | Demonstrated in India |
| Pine Labs | P3P for autonomous agentic UPI payments under permissions | Company says live in production |
| UPI Reserve Pay | Reserved funds supporting multiple eligible merchant debits | Existing NPCI capability |
| NPCI AI-agent registry | Verification and monitoring of agents interacting with UPI | Under development |
| Unified Agentic Protocol | Proposed broader framework for agentic transactions | Developing |
This status distinction is important. A completed demonstration proves technical feasibility, while a pilot tests a limited environment. A production integration can serve real users but may still have restrictions, and a proposed national protocol should not be described as universally deployed until that actually occurs.
Swiggy Is Already Bringing Commerce Into AI Assistants
Swiggy provides one of the clearest examples of how the customer journey can change. Instead of requiring a shopper to begin inside a conventional e-commerce application, Swiggy’s MCP integrations allow AI assistants to interact directly with its Food, Instamart and Dineout services.
At launch, Instamart exposed more than 40,000 SKUs through the integration. Users could use natural-language requests to search products, create carts, apply coupons and follow the delivery process through supported conversational AI tools.
This demonstrates an important change in e-commerce architecture. The merchant’s app or website is no longer necessarily the only interface through which a customer can interact with the business. An AI assistant can increasingly become another storefront, while the merchant continues to control its catalogue, inventory, order system and fulfilment.
What Is UPI Reserve Pay and Why Does It Matter?
One of the major challenges for agentic commerce is payment authorisation. AI can search and compare products easily, but allowing software to spend money creates a much higher trust requirement.
NPCI’s UPI Reserve Pay provides a mechanism through which funds can be blocked and subsequently used for multiple eligible merchant debits until the reserved amount is exhausted, revoked or expires. NPCI’s framework also requires customers to receive clear terms and notifications relating to actions such as block creation, modification, debit, revocation and expiry.
This type of infrastructure can support agentic-commerce models because the user can establish boundaries before individual purchases occur. Rather than handing unrestricted access to an AI system, the payment architecture can define how much can be spent, where transactions can occur and under what conditions.
UPI Reserve Pay itself should not be confused with a universal autonomous-shopping system. It is a payment capability on which payment providers and commerce platforms can build more sophisticated delegated or agentic transaction experiences.
Pine Labs P3P and Autonomous UPI Payments
Pine Labs announced its Pine Labs Payment Protocol, or P3P, in June 2026. The company describes it as an agentic-payment protocol designed for UPI and says it has enabled an AI agent to complete an eligible payment within predefined permissions without requiring the customer to authenticate that specific transaction at the final step.
Pine Labs describes the implementation as live in production. Claims such as “India’s first” should nevertheless remain attributed to Pine Labs because several payment companies and networks are simultaneously developing agentic-payment technologies.
The bigger significance is that agentic commerce is moving beyond product discovery. Payment providers are now attempting to solve the difficult question of how an AI agent can prove that it is acting within authority previously granted by a real customer.
Mastercard Has Demonstrated Agentic Commerce in India
Agentic payments in India are not limited to UPI. Mastercard announced in February 2026 that it had completed a fully authenticated agentic-commerce transaction on its payment network during the India AI Impact Summit in New Delhi.
The demonstration used Mastercard cards issued by Axis Bank and RBL Bank and involved payment providers including Cashfree Payments, Juspay, PayU and Razorpay. Merchant environments referenced by Mastercard included Swiggy, Instamart, Vodafone Idea, Tira and Zepto. Mastercard said the transactions were tokenised and authenticated under its Agent Pay framework.
This shows why the future of agentic commerce in India is unlikely to depend on one payment rail. UPI, cards and other authorised payment mechanisms can coexist, provided the ecosystem can establish agent identity, customer permission and transaction accountability.
NPCI’s AI-Agent Registry Could Be a Major Turning Point
One of the most important developments arrived on September 10, 2026. Reuters reported that NPCI is developing a registry designed to verify and monitor AI agents conducting transactions through UPI. The registry is expected to form part of what Reuters described as a new Unified Agentic Protocol.
The objective is to create a trusted mechanism for establishing which AI agent is attempting a transaction and whether that agent is authorised to act. Early applications are expected to focus on smaller, frequent transactions such as groceries before potentially expanding toward more complex conditional transactions.
The distinction between an agent registry and an ordinary payment API is significant. A payment system needs to know not only whether sufficient funds exist, but whether an automated actor is legitimate, whose instructions it represents and what authority it possesses.
Liability remains an important unresolved issue. If an AI agent purchases the wrong item, exceeds the customer’s instruction or makes an unauthorised transaction, the ecosystem still needs clear rules about responsibility among the consumer, AI provider, merchant, payment provider and financial institution. Reuters reports that these regulatory questions are still being developed.
Why AI-Agent Identity Matters for Merchants
For years, websites have been designed to identify and often block automated traffic. Agentic commerce creates a new problem because an automated visitor might be a malicious bot, an ordinary crawler or a legitimate AI agent representing a genuine customer.
Visa’s Trusted Agent Protocol addresses this problem by allowing approved AI agents to identify themselves cryptographically to merchants. Visa’s specifications are designed to help merchants verify the agent, understand its commerce intent and distinguish legitimate agent interactions from malicious automated traffic.
On September 10, Reuters also reported that Visa, Mastercard and Ant International are working together on common standards for identifying and verifying AI agents involved in transactions. This shows that agent identity is becoming a foundational infrastructure problem rather than something each merchant can solve independently.
How an Agentic Commerce Transaction Works
A complete agentic-commerce journey contains several layers. The user first provides an intention, such as a product requirement, budget, preferred brands, delivery deadline or spending limit. The AI system then interprets the request and searches permitted data sources or merchant systems for suitable products.
The agent may compare product attributes, pricing, delivery charges, availability and policies before selecting a suitable option. If action is permitted, it can interact with the merchant’s commerce interface to create a cart or prepare an order. A trust and authorisation layer then determines whether the agent is genuine and whether the requested action falls within the user’s permission.
The final transaction can then move through a suitable payment mechanism. After payment, normal commerce functions such as order confirmation, delivery tracking, cancellation, returns and customer support still need to work. This is why agentic commerce is not simply an AI-model feature; it connects AI with product data, merchant APIs, identity, payments and fulfilment.
ACP, UCP, MCP and AP2: What Businesses Need to Know
The agentic-commerce ecosystem is developing several protocols. Businesses do not need to implement every protocol, but understanding their different roles is useful because no single standard currently controls the entire AI-commerce journey.
| Protocol | Main Role | Why It Matters to Commerce |
|---|---|---|
| MCP | Connects AI systems with tools and services | Allows an AI assistant to interact with merchant functions such as product search and cart operations |
| ACP | Agentic Commerce Protocol | Supports AI-native product discovery and commerce interactions |
| UCP | Universal Commerce Protocol | Provides a common commerce language between AI surfaces, merchants and payment providers |
| AP2 | Agent Payments Protocol | Focuses on secure authorisation and payment by agents |
OpenAI expanded its Agentic Commerce Protocol in March 2026 to support richer product discovery in ChatGPT. Merchants can provide more complete product information such as pricing, reviews and features, allowing users to compare products directly within the AI experience.
Google introduced the open-source Universal Commerce Protocol in January 2026. UCP is designed to connect consumer AI surfaces with businesses and payment providers while supporting integration through APIs, Agent2Agent and MCP. Google developed it with companies including Shopify, Etsy, Wayfair, Target and Walmart, with ecosystem support that includes Flipkart and major payment providers.
Google’s Agent Payments Protocol addresses payment authority. AP2 v0.2 added support for “Human Not Present” transactions, in which an AI agent can make an authorised purchase later according to instructions established by the customer in advance. Google transferred AP2 to the FIDO Alliance in April 2026 to support a more platform-neutral standard.
Agentic Commerce Does Not Mean Giving AI Your UPI PIN
A secure agentic-payment architecture should not require consumers to type raw UPI PINs, card credentials or other sensitive authentication secrets into an AI conversation. The objective of emerging payment systems is to separate the AI agent from the sensitive credentials used to authorise and settle a transaction.
That can be achieved through mechanisms such as payment tokenisation, pre-authorised spending boundaries, cryptographically signed instructions and revocable permissions. Mastercard’s India demonstration, for example, used tokenised and authenticated transactions rather than simply exposing payment credentials to an AI model.
For businesses implementing AI-shopping functionality, this distinction is fundamental. The AI agent should receive only the information and permissions required to perform its task, while regulated payment infrastructure should continue handling sensitive authentication and settlement functions.
Which Indian Businesses Can Benefit Most From Agentic Commerce?
The strongest early use cases are likely to involve products or services that are purchased frequently, have structured inventory and can be described through clear constraints. Grocery, quick commerce, food delivery, household replenishment, subscriptions, beauty products, routine travel and certain B2B procurement journeys fit this pattern particularly well.
A consumer buying the same household items every week can safely define useful limits such as brand preferences, quantity, maximum spend and acceptable substitutions. That is easier to automate than a high-value, highly subjective or regulated purchase requiring extensive professional judgement.
The same principle applies in B2B commerce. A business could eventually authorise an agent to reorder approved office supplies when inventory drops below a threshold, provided vendors, quantities, prices and spending limits are clearly defined. The value comes from eliminating repetitive purchasing work without removing appropriate controls.
What Agentic Commerce Means for E-commerce Websites
Agentic commerce does not mean e-commerce websites will disappear. It means the website or app may no longer be the only interface through which a customer reaches the merchant.
A future merchant could simultaneously serve people browsing a website, Google and other search crawlers, AI systems researching products, shopping assistants retrieving product information and authorised agents attempting to create orders. The merchant therefore needs both a strong human experience and a reliable machine-readable commerce layer.
For businesses planning new stores or modernising an existing platform, e-commerce development increasingly needs to consider catalogue architecture, structured product data, APIs, inventory availability and integrations in addition to frontend design and checkout UX.
How Agentic Commerce Changes SEO and AI Visibility
Traditional e-commerce SEO tries to ensure a product can be discovered by a search engine and then persuade a person to visit the website. Agentic commerce adds another layer: the business may also need an AI system to understand the product well enough to retrieve, compare and potentially recommend it.
This makes accurate product information increasingly important. Product names, prices, currency, availability, variants, identifiers, specifications, shipping information, returns, ratings and warranties should be consistent and machine-readable wherever technically possible.
Schema markup remains useful because it reduces ambiguity around entities such as products, offers, organisations, ratings and policies. However, structured data alone does not guarantee inclusion or recommendation inside ChatGPT, Gemini or another AI shopping system. Businesses also need accurate feeds, reliable merchant information, strong brand/entity signals and, where required, compatible commerce integrations.
This is where traditional SEO begins to overlap with Generative Engine Optimization and LLM optimization. The objective is no longer merely to repeat keywords; it is to make the business and its products understandable, retrievable and trustworthy across search engines and AI-driven interfaces.
What Information Should an Agent Be Able to Understand About a Product?
| Information | Why It Matters |
|---|---|
| Canonical product name | Prevents confusion between similar items |
| Brand and identifier | Helps match the exact product |
| Current price and currency | Allows accurate comparison |
| Inventory status | Prevents recommendation of unavailable products |
| Variants | Clarifies size, colour, capacity or configuration |
| Delivery information | Allows agents to satisfy time-based requests |
| Shipping cost | Enables comparison of total purchase cost |
| Return policy | Helps evaluate purchase risk |
| Warranty | Provides post-purchase confidence |
| Specifications | Allows constraint-based product matching |
| Reviews and ratings | Provide useful customer evidence when genuine |
If crucial information exists only in promotional imagery or cannot be retrieved reliably from the merchant’s systems, an AI agent has less dependable information on which to make a recommendation. Improving product data therefore becomes both an e-commerce-quality issue and an AI-commerce-readiness issue.
What Businesses Need to Become Agentic-Commerce Ready
Most businesses do not need to start by building a fully autonomous shopping agent. A better first goal is to become agent-ready: create commerce infrastructure that authorised AI systems can understand and interact with safely when the opportunity becomes commercially relevant.
| Area | Business Requirement |
|---|---|
| Product catalogue | Accurate and consistently structured products, variants and identifiers |
| Inventory | Current stock information accessible to permitted systems |
| Pricing | Reliable prices, taxes, discounts and fees |
| Policies | Clear delivery, cancellation, return and refund terms |
| Structured data | Valid markup for relevant products, offers and business entities |
| Commerce APIs | Secure mechanisms for approved systems to search, create carts and place orders |
| Agent identity | Ability to distinguish authorised AI agents from malicious bots |
| Payment authority | Clear, limited and revocable customer permission |
| Logging | Auditable records of significant agent actions |
| Post-purchase systems | Order tracking, cancellation, refunds and support |
| Human escalation | A way to stop automation when a situation is uncertain or high risk |
A Practical Agentic Commerce Roadmap for Businesses
The first stage should focus on data quality. Audit the catalogue, remove inconsistent product names, standardise variants, confirm prices and stock, document shipping and returns, review structured data and identify where important commerce information currently exists only inside webpages rather than reliable systems.
The second stage should test one narrow customer journey. A business could allow an AI interface to understand a product request, search its catalogue and create a cart while keeping final checkout under direct customer control. Measure whether the correct products are retrieved, how often customers need to correct the agent and where failures occur.
The third stage can introduce carefully controlled transactions after the discovery and cart stages have proven reliable. Spending limits, category restrictions, authentication, audit trails, revocation, exception handling and refund processes should be defined before greater autonomy is enabled.
Companies that need multi-step AI workflows can connect these systems through AI automation services and suitable workflow infrastructure such as n8n automation. Businesses requiring their own customer-facing AI system may instead need custom AI application development.
How to Measure Agentic Commerce Performance
Businesses should not judge success by the number of AI conversations alone. Agentic commerce ultimately needs to improve the economics or customer experience of a real purchasing journey.
| Metric | What It Measures |
|---|---|
| Product retrieval accuracy | How often the agent identifies products matching the request |
| Recommendation-to-cart rate | Whether recommendations are useful enough to progress |
| Cart-to-order conversion | Commercial effectiveness of the journey |
| Human intervention rate | How much manual help is still required |
| Incorrect-order rate | Whether automation is creating purchasing mistakes |
| Payment success rate | Reliability of the payment stage |
| Refund/dispute rate | Whether purchases are creating downstream problems |
| Average order value | Commercial value of agent-driven transactions |
| Repeat purchase rate | Whether customers find the experience useful enough to reuse |
| Cost per agentic order | AI, infrastructure and transaction cost required per order |
| Incremental revenue | Sales that would not otherwise have occurred |
The strongest agentic-commerce system is therefore not the one that performs the most autonomous actions. It is the system that produces accurate purchases, high customer trust, sustainable economics and fewer unnecessary steps.
What Are the Main Risks of Agentic Commerce?
Allowing AI to move from recommendations to financial actions creates materially higher risk. The agent can misunderstand the user’s request, select the wrong variant, act on outdated inventory, encounter manipulated content or execute an action outside the customer’s intended boundaries.
Merchants also need protection from fraudulent agents pretending to represent customers. This explains the growing focus on cryptographic agent identity from companies such as Visa and on NPCI’s planned agent registry for UPI.
Payment authority should therefore be narrow, explicit and revocable. An AI agent authorised to replenish groceries up to a specified amount should not automatically inherit permission to make unrelated high-value purchases.
Businesses should also preserve a reliable audit trail showing what the user requested, what the agent selected, what information the merchant supplied, what transaction was authorised and what actually happened. This becomes particularly important when customers dispute an AI-initiated purchase.
Can Agentic Commerce Reduce Website Traffic?
Yes, at least for some purchase journeys. If an AI assistant can discover products, compare them and create an order without requiring the consumer to browse the merchant’s website, traditional website sessions may no longer capture the merchant’s full commercial influence.
This does not necessarily mean fewer sales. A merchant could receive more transactions while recording fewer conventional product-page visits. Analytics may therefore need to evolve from measuring only clicks and sessions toward tracking AI referrals, product-data retrieval, agent-created carts, API interactions and completed transactions.
For marketers, this makes e-commerce marketing more closely connected with technical product feeds, server-side attribution, structured data and AI-search visibility.
Could AI Agents Change How Products Compete?
AI agents can evaluate considerably more information than a typical customer is willing to compare manually. A consumer might compare three stores before becoming tired of researching, whereas an agent could potentially evaluate many merchants according to price, shipping, ratings, returns and product specifications.
This can increase price transparency, but the cheapest product will not necessarily always win. Delivery reliability, merchant reputation, warranty, return policy, product quality, verified reviews and brand authority can all become important decision signals.
For brands, the implication is that good marketing cannot consist only of attractive creative. Product claims need supporting evidence, commercial conditions need to be clear and the business needs enough trustworthy information for an AI system to distinguish it from inferior alternatives.
Will Agentic Commerce Replace E-commerce Websites?
Agentic commerce is unlikely to eliminate websites or mobile apps in the near term. Human customers will continue to want visual exploration, product storytelling, customer support, brand experiences and direct control over many important purchases.
What is more likely is a multi-interface model. The same merchant may sell through its website, mobile app, marketplace listings, search engines, social platforms and AI assistants. Agentic commerce becomes another access layer rather than an immediate replacement for every existing channel.
What Should Indian Businesses Do Now?
Businesses do not need to wait for fully autonomous AI shopping before preparing. The most valuable work can begin with improvements that already benefit ordinary e-commerce and search visibility: accurate product data, current inventory, transparent policies, valid structured data, reliable merchant information and well-designed APIs.
Companies with high transaction volume should then identify one repetitive, low-risk customer journey where AI could remove meaningful friction. Build and measure that workflow before extending autonomy to more products or higher-value transactions.
Payment automation should come after product selection and order creation are reliable, not before. A technically impressive AI checkout system has little value if the agent regularly chooses the wrong product or cannot understand the merchant’s return conditions.
Frequently Asked Questions About Agentic Commerce in India
What is agentic commerce?
Agentic commerce is a model in which an AI agent can perform commercial actions on behalf of a user or business, such as discovering products, comparing alternatives, building a cart and, where appropriately authorised, initiating or completing a transaction.
Is agentic commerce available in India?
Yes, selected agentic-commerce integrations and payment implementations already exist in India. Swiggy has AI-native commerce integrations, Mastercard has demonstrated authenticated transactions, Pine Labs says its P3P protocol is in production, and NPCI is developing broader agent-verification infrastructure. However, universal autonomous shopping across all Indian merchants is not yet available.
Can AI agents make UPI payments?
Agent-mediated UPI payment implementations are developing in India. UPI Reserve Pay provides useful pre-authorised payment capabilities, Pine Labs says its P3P protocol enables qualifying autonomous UPI transactions, and NPCI is developing additional infrastructure for verified AI agents. These developments should not be interpreted as allowing any arbitrary AI agent unrestricted access to a person’s UPI account.
What is the Unified Agentic Protocol in India?
Reuters reported on September 10, 2026 that NPCI is developing a Unified Agentic Protocol that includes a registry for authenticating and monitoring AI agents operating through UPI. The framework remains under development and should not yet be described as a universally deployed payment standard.
Is agentic commerce the same as conversational commerce?
No. Conversational commerce means interacting with a merchant through chat or voice. Agentic commerce adds the ability for AI software to perform actions. A conversational interface can therefore be agentic, but simply answering product questions does not make a chatbot an agentic-commerce system.
Is agentic commerce the same as AI shopping?
Not always. AI shopping can include simple product discovery and recommendations. Agentic commerce generally implies that the AI can interact with commerce systems and take permitted actions rather than only producing advice.
Does an AI shopping agent need my UPI PIN?
A properly designed agentic-payment system should not require users to expose raw UPI PINs inside AI prompts. Emerging systems use approaches such as tokenisation, pre-authorised permissions and specialised payment infrastructure so sensitive credentials remain separated from the language model.
How will agentic commerce affect SEO?
SEO will remain important for discovery, but merchants increasingly need structured, accurate product information that AI systems can understand and retrieve. Product feeds, schema markup, entity authority, current pricing, inventory and machine-readable policies can become increasingly important alongside conventional rankings.
Does my business need an MCP server for agentic commerce?
Not necessarily. MCP is one method for connecting AI systems with merchant tools. Depending on the platform, businesses may use existing APIs, MCP, ACP, UCP or integrations provided by commerce and payment partners. The underlying requirement is that authorised systems can reliably retrieve information and perform controlled commerce actions.
Which businesses are best suited to agentic commerce?
Businesses with high-frequency purchases, structured inventory and predictable decision rules are among the strongest early candidates. Grocery, food delivery, quick commerce, replenishment products, subscriptions and certain B2B procurement workflows are easier to automate safely than highly subjective or regulated purchases.
