The agentic evolution of eCommerce: From generative AI to autonomous agents

You don’t have to be a technology expert to know — or at least sense — how much generative AI has transformed our lives. From content creation to new ways of searching the web, GenAI has become a universal tool, causing a mix of excitement, curiosity and, quite often, controversy. 

Businesses quickly seized the opportunity to leverage GenAI’s powerful capabilities, scaling creativity and engaging customers more effectively by generating product descriptions, personalizing recommendations and even creating dynamic pricing suggestions. 

Generative AI covers a range of models — text, images, audio and more — with LLMs (large language models) being one specific category. It excels at creating, personalizing and even surfacing content based on content that already exists. It creates new content based on what it ingests, but it remains primarily reactive, responding to prompts or inputs. 

The next frontier in artificial intelligence is agentic AI: Systems that can plan, decide and execute multi-step workflows autonomously. These agents can act on behalf of humans or businesses, completing entire shopping journeys without (or with minimal) human intervention. 

So, what’s the common denominator? And how do generative and agentic AI intersect? 

Consumers are quickly adopting GenAI channels, such as ChatGPT, Perplexity, Microsoft Copilot and Google’s Gemini, for online shopping. These tools are fundamentally generative: They help shoppers explore options, compare products and receive personalized recommendations in natural language. 

But what’s emerging now is that these same generative interfaces are becoming gateways for agentic AI — systems capable of taking the next step from suggestion to action. In more detail:

• Generative AI serves as the conversational and creative layer. It understands intent, generates options and frames responses in a human-friendly way. 

• Agentic AI, in turn, provides the execution and orchestration layer: Once intent is clear, it plans, decides and performs actions such as checking stock, applying discounts or placing orders. 

In short, this intersection is what transforms commerce from a series of manual decisions into an intelligent, end-to-end experience where discovery, decision and delivery flow together.

Generative and agentic AI complement each other. Agentic AI interprets that information, makes decisions and takes autonomous actions within it. Generative AI creates responses based on personalized and structured information.

Example: How an AI Agent completes an end-to-end shopping journey

Customer need: Restock a favorite pair of running shoes.

• Agentic AI: Understands the intent, researches options, compares prices, checks stock, talks to seller APIs, places the order and tracks the shipment.

• Generative AI: Communicates with the customer, explains choices and updates on progress in natural language.

As agentic AI begins to take action on behalf of consumers, the commerce ecosystem needs a common language and set of standards that allow autonomous agents to interact safely and effectively with online stores, platforms and services. 

For enterprises, this means mastering and deploying emerging integration enablers, including MCP, A2A, AP2 and ACP. 

What is MCP?

Anthropic’s MCP (Model Context Protocol) is an open framework that allows AI models, such as ChatGPT, Gemini or Claude, to securely connect with external data sources, tools and APIs. 

It enables agents to access live, contextual information beyond their training data while respecting privacy and permission boundaries. In practical terms, MCP acts as the bridge between large language models and the real-world systems they interact with, allowing them to query databases, execute tasks, or retrieve up-to-date information safely and consistently.

What is A2A?

A2A (Agent-to-Agent Protocol) defines how autonomous agents communicate, collaborate and negotiate with each other. 

As agent ecosystems expand, A2A creates a common language for inter-agent coordination, enabling different AI agents (e.g., a shopping agent, a logistics agent, and a payment agent) to exchange data, delegate tasks and resolve actions without human oversight.

This standard helps ensure that agent interactions remain transparent, auditable and aligned with user intent, opening the door to more complex, multi-agent commerce workflows where cooperation replaces isolation.

What is AP2? 

AP2 is Google’s Agent Payments Protocol, a standard designed to enable autonomous and semi-autonomous AI agents to make verifiable purchases on behalf of users. AP2 ensures that autonomous transactions follow trusted payment flows and include:

• Verified consent.

• Tokenized authorizations.

• Fraud controls.

• Support for subscriptions, one-time purchases and split payments.

What is ACP? 

ACP (Agentic Commerce Protocol) is an open standard designed to help AI agents securely connect with commerce systems, complete purchases and sync product catalogs, giving enterprises a universal path to participate in agent-driven commerce on ChatGPT. Pioneered by OpenAI, the emerging standard was created in partnership with Stripe and commercetools.

In other words, instead of each agent learning to navigate every unique checkout flow, website structure or API format, ACP defines a consistent way for agents to:

• Discover and interpret product data, pricing and availability.

• Verify trust and compliance signals.

• Execute actions like “add to cart,” “apply discount,” “place order” and “track shipment.”

• Communicate outcomes or exceptions back to users in a standardized way.

ACP reduces the friction between AI agents and commerce systems, paving the way for a future where autonomous shopping can happen in a secure and transparent fashion on ChatGPT.  

Building agentic capabilities from scratch means integrating each GenAI channel through specific protocols, securing APIs, managing data flows and enforcing compliance — before you can even think about the first transaction. 

For enterprises to take advantage of agentic commerce, commercetools simplifies this with two core capabilities: AI Hub and Agent Gateway.

AI Hub: Connects your product catalog, pricing and checkout to emerging channels like ChatGPT (using ACP), Microsoft Copilot and Perplexity (coming soon), making your brand instantly discoverable and shoppable. commercetools unifies data and instant checkout in real-time, keeping your brand compliant with every new agentic standard and protocol. 

Agent Gateway: A fully managed service layer ensuring AI agents interact with your commerce systems securely and predictably through authentication, permission boundaries and continuous auditing. 

Together, these form a single, unified commerce data layer that gives every agent a consistent understanding of products, prices and orders across channels. 

This agentic foundation comes pre-engineered and production-ready for enterprises, cutting implementation time from months to weeks. 

Generative AI channels are poised to become the primary interface for commerce, ushering in a new era of “zero-click” shopping where traditional websites take a back seat. For instance, traffic to US retail sites has surged by 4,700% via AI-powered channels, and over 60% of consumers have used conversational AI for purchases. 

But the landscape is fragmented and complex. Enterprises require a secure, interoperable foundation to quickly integrate with any LLM.

A package combining AI Hub + Agent Gateway, commercetools Agentic Jumpstart, enables enterprises to:

• Connect to leading GenAI channels.

• Convert AI-driven intent into transactions.

• Rapidly build agentic commerce flows.

• Ensure compliance and security.

With Agentic Jumpstart, enterprises have a launchpad to reach AI-driven shoppers — and turn their intent into conversion — starting now. 

Don’t navigate the AI momentum alone. Contact our experts to start your agentic commerce now. 

How does agentic AI differ from generative AI in the context of shopping?

Generative AI provides conversational guidance and frames responses in natural language, while agentic AI takes autonomous actions such as placing orders, applying discounts or coordinating with merchant APIs.

What protocols enable agentic commerce?

Key protocols include MCP (Model Context Protocol), A2A (Agent-to-Agent Protocol), AP2 (Agent Payments Protocol) and ACP (Agentic Commerce Protocol), which allow agents to fetch data, collaborate and securely complete transactions.

Why are enterprise retailers adopting agentic commerce now?

Retailers are turning to agentic commerce to meet shoppers on emerging GenAI channels, reduce friction in checkout flows, enable “zero-click” shopping and boost conversion through autonomous, personalized assistance.

How does commercetools support agentic commerce?

commercetools two add-ons — AI Hub and Agent Gateway — through the package Agentic Jumpstart, giving enterprises a unified, secure foundation to connect GenAI channels and enable autonomous shopping across multiple LLMs.