You open your laptop to start a routine set of tasks. You need to restock supplies, book a vendor meeting, and source a new supplier. In the traditional world, that takes hours. Browser tabs start piling up. You compare prices manually. You fill out the same form fields on three different checkout pages.
Now imagine an AI agent handling all of it in minutes, with accuracy and no manual effort. This is not a distant future. It is agentic commerce in action today.
Agentic commerce is the use of autonomous AI agents to research, compare, negotiate, and complete purchases on behalf of users or businesses, often without any direct human intervention. It goes far beyond chatbots or recommendation engines. It fundamentally redefines how buying and selling decisions are made.
By 2030, the US B2C retail market alone could see up to $1 trillion in orchestrated revenue from agentic commerce. Globally, that number climbs to between $3 trillion to $5 trillion, according to McKinsey research. And Gartner projects that 33% of enterprise software applications will include agentic AI by 2028, up from less than 1% in 2024.
If your business is still building for human browsers alone, the window to adapt is closing fast.
In this blog, you will learn what agentic commerce is, how AI agents move from user intent to completed purchase, and what protocols like MCP, A2A, and ACP are making it all possible. You will also get the six business domains every enterprise must evolve, a four-step preparation roadmap, and a look at how SPXCommerce fits into an agent-ready commerce stack.
What Exactly Is Agentic Commerce?
Agentic commerce is where AI moves from assisting to acting.
Earlier generations of retail AI were reactive. Recommendation engines surfaced products. Chatbots answered scripted questions. Predictive tools flagged inventory issues. All of these systems required constant human prompting to move forward.
Agentic AI is different in three fundamental ways.
Autonomy
Agents act without constant user input. They follow predefined frameworks and guardrails, executing multi-step workflows from a single high-level instruction.
Reasoning
Agents adapt in real time to changing conditions and constraints. If a price changes, a product sells out, or a shipping deadline shifts, the agent adjusts its decision without looping a human back in.
Interoperability
Agents integrate across platforms, APIs, and systems to execute end-to-end workflows. They can query a product catalog, validate inventory, apply a discount code, and complete a checkout across multiple vendors in one flow.
As Salesforce CEO Marc Benioff put it: “They can perform tasks independently, make decisions and even negotiate with other agents on our behalf.”
This is the core of agentic commerce. The agent is not just a helper. In many cases, the agent is the buyer.
How Agentic Commerce Actually Works
Understanding how this works is critical for adoption. Agentic commerce flows through several interconnected stages.
1. User-to-Agent Engagement
It starts with a human intent. A consumer says, “Find me a waterproof hiking boot under $150 that arrives by Friday.” A procurement manager says, “Source three approved vendors for industrial fasteners and get quotes.”
The agent interprets that intent. It applies the user’s constraints, including budget, preferences, delivery windows, and brand restrictions, and gets to work. Done well, this feels less like issuing a command and more like briefing a trusted colleague.
2. Autonomous Execution
Once activated, the agent plans a multi-step workflow and executes it independently. It queries product catalogs via structured APIs. It compares options in real time. It applies coupons, checks return policies, and verifies availability across multiple retailers simultaneously.
For low-risk purchases, this can be fully automated. For high-value or sensitive transactions, the agent flags a decision point and waits for human approval before proceeding.
3. Product Discovery and Decision-Making
This is where agentic commerce diverges most sharply from traditional search. The agent does not browse. It operates with a clear goal. It analyzes structured product data from multiple sources and ranks options against the user’s exact criteria, including price, delivery time, reviews, return policy, and availability.
This is also why Generative Engine Optimization (GEO) is emerging as critical for visibility in AI-driven discovery.
Brands now need machine-readable product data, standardized attributes, and clear metadata, not just for human search engines, but for the AI systems acting as buyers.
4. Merchant-to-Agent Interaction
For agents to transact at scale, merchants must expose their systems through machine-readable interfaces. This means structured APIs for product catalogs, real-time pricing, inventory, and return policies.
This is driving interest in the Agentic Commerce Protocol (ACP), an emerging standard for how AI agents and merchants exchange structured information. OpenAI and Stripe have co-developed an early version of this, allowing users to complete purchases directly within ChatGPT.
5. Secure Agentic Payments
Agentic payments form the critical infrastructure that enables secure autonomous transactions. Without secure, auditable transaction rails, autonomous purchasing cannot scale.
Major platforms are moving fast here. Google’s Agent Payments Protocol (AP2) uses cryptographically signed mandates to link user intent, cart, and payment, creating an auditable trail that cannot be repudiated. Visa is piloting AI-ready tokenized credentials in partnership with Anthropic, Microsoft, OpenAI, and Stripe. Mastercard is developing its Agent Pay solution.
Stripe, which powers 78% of the Forbes AI 50, has built agentic payment flows specifically for agents that are authorized to spend but must operate on secure, auditable rails. Over 700 AI agent startups launched on Stripe in 2024 alone.
6. Post-Purchase Support
The agent’s job does not end at checkout. It can track shipments, manage returns, flag delivery issues, and initiate after-market recommendations for complementary products, all without requiring the user to log in anywhere.
The Five Attributes Every Enterprise Agent Needs
Building effective agentic commerce infrastructure requires clarity on what agents are, what they can do, and what they must never do.
Role
Define the job in natural language. What is the agent responsible for? A merchandising agent tasked with liquidating slow-moving inventory needs a different scope than a procurement agent sourcing new suppliers.
Data
An agent is only as good as the data it can access. Fragmented, inconsistent product data breaks agentic workflows. Clean, structured, unified data across inventory, pricing, fulfillment, and customer history is the foundation. Businesses should prioritize structured formats like schema.org markup and GS1 standards so that AI systems can interpret offerings without ambiguity.
Actions
Define the exact tasks the agent can execute. Each action should run on robust, API-driven workflows designed for speed and cross-platform interoperability. A single agent might identify excess inventory, cross-reference it with high-value customer segments, generate a targeted promotion, and push it live across all digital storefronts without a single human touchpoint.
Guardrails
Define what the agent must never do. These can be natural-language restrictions, such as always escalating purchases above $10,000 to a human, or built-in security frameworks that limit scope, flag anomalies, and enforce compliance policies.
Channels
Agents work where your business operates, including your website, your CRM, WhatsApp, Slack, and your mobile app. The channel defines the context. A shopper-facing agent on a retail site behaves differently from an internal procurement agent embedded in an enterprise workflow tool.
What This Means for Enterprise Commerce Teams
Agentic commerce is not limited to consumers. It is a full enterprise transformation.
1. Merchandising and Marketing
Consider what a merchandising team spends hours doing today: pulling sales performance by category, analyzing campaign engagement, identifying slow-moving SKUs, and designing promotional strategies. That entire process can be automated by a single agent running on a recurring schedule.
An apparel retailer might instruct an agent: “Every Monday, identify the three lowest-performing items in each category and suggest targeted promotions for our most likely-to-buy segments.” The agent pulls the data, runs the analysis, drafts the promotions, and surfaces recommendations before the team has finished their first coffee.
Agents can also write and refresh product descriptions at scale. They can pull signals from customer reviews to proactively address common questions in listings. Something like “Most customers say this jacket runs large. Size down for a more tailored fit” can be generated and published without any manual editing required.
2. B2B Procurement and Supply Chain
This is where the enterprise stakes get very high. Agentic commerce allows procurement teams to automate routine purchasing decisions entirely. Agents can validate approved vendors, check contract compliance, negotiate volume-based pricing, and place orders within preset parameters.
61% of procurement leaders now cite geopolitical and supply chain risks as top concerns. By 2028, half of G2000 manufacturers are expected to operationalize AI-enabled circular supply chains. Agentic commerce is the coordination and transaction layer that makes that vision operational.
3. Travel and Hospitality
End-to-end booking workflows, including flights, hotels, ground transportation, and event tickets, are a natural fit for agentic automation. Agents can monitor prices, rebook automatically when conditions change, and handle refunds or credits within predefined approval limits.
4. Digital Subscription Management
Agents can monitor usage, cancel underperforming subscriptions, upgrade plans when thresholds are hit, and switch providers based on performance or price changes. This is post-purchase optimization at a scale no human team can match.
The Protocols Powering Agentic Commerce Infrastructure
This is not vaporware. The technical infrastructure for agentic commerce is being built and deployed right now.
Model Context Protocol (MCP)
Developed by Anthropic, MCP is an interoperability standard that allows AI agents to share context, intent, and memory across models and tools. Instead of making isolated API calls, agents using MCP retain reasoning and objectives across environments. This is the foundation for coherent, autonomous behavior at scale.
Agent-to-Agent Protocol (A2A)
A2A enables autonomous agents, regardless of vendor, architecture, or environment, to coordinate directly with each other. Built on JSON-RPC and HTTP, it supports long-running tasks, dynamic capability discovery, and multi-modal collaboration. This is what makes multi-agent ecosystems possible in real-time commercial environments.
Agent Payments Protocol (AP2)
Google’s AP2 is an open, payment-agnostic protocol that enables agents to make verifiable purchases on behalf of users. Cryptographically signed mandates link intent, cart, and payment in an auditable chain. It is backed by Mastercard, PayPal, American Express, Adobe, and Alibaba.
Agentic Commerce Protocol (ACP)
ACP governs how agents and merchants exchange structured information, including product data, pricing, availability, and policies, at machine speed and scale.
The duration of tasks that LLMs can reliably complete has been doubling every seven months since 2019 (METR). Anthropic’s Claude 4.5 can now complete tasks requiring more than 30 hours of skilled human effort. The infrastructure is catching up to the capability.
Why Consumers Still Have Concerns and What That Means for Your Business
Trust is the biggest barrier to agentic commerce adoption. IBM research shows that 83% of consumers share overlapping concerns about privacy, data misuse, and unsolicited marketing in AI-driven purchase flows.
These concerns are valid. When a human makes a purchase, trust is visible. When an agent makes it, trust becomes abstract, filtered through automation, data pipelines, and institutional frameworks. Consumers are right to ask: Who is accountable if something goes wrong?
For enterprise leaders, this is not a technology problem. It is a governance problem.
Businesses that build trust into their agentic infrastructure will have a durable competitive advantage. That means deploying clear audit trails for every agent-initiated transaction, explainable decision logs that users can review, tiered autonomy with more human oversight for high-risk purchases and full automation for low-risk ones, transparent data policies that agents surface proactively, and easy human override at every stage.
Consent cannot be a checkbox. It must be a dynamic, ongoing agreement that users can adjust as their comfort level evolves.
Six Business Domains That Must Evolve Now
McKinsey identifies six key domains every retail business must address to compete in the agentic era. Three require full innovation. Three require strategic renovation.
Innovation Domains
Customer Engagement and Product Discovery. Build agents that understand intent and proactively surface products, bundles, and alternatives. Embed semantic and behavioral metadata into catalogs. Develop agent-authenticated interfaces for autonomous discovery.
Clienteling and Loyalty. Create hyperpersonalized experiences triggered by inferred intent. Build persistent customer-context layers accessible by agents. Expose loyalty APIs so agents can apply points, rewards, and eligibility in real time.
Payments and Fraud Detection. Shift from stopping bots to enabling trusted agents. Build programmable spend policies, delegated authorization systems, and fraud models designed for agent behavior, not human behavior.
Renovation Domains
Core Commerce Platforms. Enable agents to execute structured transactions with minimal human input. Integrate dynamic pricing, inventory-aware recommendations, and agent-compatible checkout flows.
In-Store Point of Service. Synchronize digital and physical journeys by sharing context with store associates, integrating spatial computing, and providing digitized inventory maps accessible to agents.
Fulfillment and Returns. Deploy agents to automate fulfillment decisions, negotiate return logic, and orchestrate post-purchase actions through agent-ready APIs and modular carrier integrations.
How to Prepare Your Business for Agentic Commerce
Agentic commerce rewards the most prepared, not the fastest spenders. The businesses pulling ahead today made deliberate infrastructure decisions early: clean data, open APIs, structured catalogs, and governance that scales. These four steps are the operational foundations that determine whether AI agents can find you, evaluate you, and buy from you.
Step 1: Standardize Your Product Data
If your product data is fragmented across systems or presented in inconsistent formats, your agents cannot function accurately. Unify product, pricing, inventory, fulfillment, and customer data into a single, structured format. Implement schema.org markup and GS1 standards. Make your catalog machine-readable, not just human-readable.
Step 2: Build and Expose Open APIs
Agentic commerce runs on API-first infrastructure. Merchants who expose clean, well-documented APIs for product data, real-time availability, pricing, and policies will be discoverable to agents. Those who do not will be invisible, and invisible means out of the consideration set entirely.
Step 3: Rethink SEO for AI Discovery
Traditional SEO optimizes content for human search behavior. Agentic commerce requires Generative Engine Optimization (GEO), which structures product content so that LLMs and AI agents can interpret, rank, and act on it. As 44% of users who have tried AI-powered search say it has become their primary search method (McKinsey), the stakes for GEO are rising fast.
Step 4: Define Clear Guardrails and Governance
Before you deploy any agentic capability, define exactly what your agents can and cannot do. Set spend limits. Require human approval for high-risk decisions. Build audit logs. Establish escalation paths. Trust is earned through consistent, transparent behavior, not just built at launch.
The Competitive Reality: First-Movers Will Define the Rules
Perplexity launched its agentic shopping tool “Buy with Pro” in late 2024. OpenAI’s Operator, launched in January 2025, enables agents to book travel and complete purchases within ChatGPT. Shopify is building agentic infrastructure that lets agents build carts across merchants. Amazon, Google, PayPal, and Mastercard are all developing agentic shopping services.
The competitive window is not infinite.
Early adopters of agentic commerce will do more than gain market share. They will influence how AI systems handle discovery, recommendation, and loyalty across the entire ecosystem. They will set pricing models before standards are locked in. They will capture consumer intent before competitors even have agents in market.
As McKinsey senior partner Lareina Yee put it: “The companies that move first, even in small ways, will be the ones that help shape the future.”
Half of all consumers already use AI during some part of their buying journey (IBM Institute for Business Value, 2026). That number will only climb.
How SPXCommerce Positions Your Business for Agentic Commerce
Most commerce platforms were built for a world where humans do the browsing. SPXCommerce was built for what comes next. As an AI-first, enterprise-grade ecommerce platform, SPXCommerce gives brands the infrastructure they need to compete in an agent-driven marketplace. Its microservices architecture delivers sub-2-second page load speeds, while its AI-powered suite handles conversational analytics, predictive demand forecasting, smart vendor onboarding, and multi-lingual content generation out of the box. These are not add-ons bolted onto a legacy system. They are core to how the platform was engineered from day one.
What makes SPXCommerce directly relevant to agentic commerce is its foundation in structured, machine-readable data and API-driven operations. Its centralized Product Information Management (PIM) system keeps catalog data clean, consistent, and syndication-ready across every channel, which is exactly the data quality that AI agents require to discover, evaluate, and transact with confidence. Combined with its Business Intelligence AI that lets teams query performance data in plain language, SPXCommerce positions businesses to be both agent-ready and human-ready at the same time, so that no matter who is doing the buying, your brand stays in the consideration set.
The Bottom Line
Agentic commerce is not a future trend to monitor. It is an active market shift with quantifiable economic stakes and infrastructure already being deployed by the largest technology companies in the world.
For enterprise leaders, the question is no longer whether AI agents will reshape online marketplaces. They already are. The question is whether your business will be discoverable, transactable, and trustworthy in a world where the buyer might not be human.
That means clean data. Open APIs. Agent-compatible checkout flows. Governance frameworks built for autonomous transactions. And a willingness to rethink business models that were designed for a browser-and-click world.
The businesses that act now will not just survive this shift. They will define what commerce looks like on the other side of it.
