Artificial Intelligence E-commerce

How Agentic Commerce Is Revolutionizing E-commerce (and Why to Act Now)

on 3 November 2025 9 minutes read

Or summarize with: ChatGPT Perplexity Claude Mistral

What is agentic commerce? How should merchants approach it? What will it change in online retail? You may feel a little lost and be looking for concrete answers to make the right decisions and get your business ready. Take a few minutes to read this analysis.

What is agentic commerce?

Agentic commerce is what happens when the upheaval brought about by artificial intelligence in recent years reaches online retail, driven in particular by conversational agents such as Claude, Gemini and ChatGPT.

From the traditional journey to agentic commerce: an overview

Until now, in a traditional journey, shoppers typed a query into Google, got a list of matching products, and were sent to a classic e-commerce site to complete their purchase.

Today, with agentic commerce, shoppers talk with an AI to describe exactly what they need. The AI does all the work of aggregating the available content and suggests the most relevant products right in the conversation. Shoppers interact with the AI to refine what they want, then choose a product and complete the transaction directly in the conversational agent’s interface.

Yesterday’s and today’s processes compared:

Process step	Traditional process (yesterday)	Agentic process (today)
1. Intent & Search	The user starts the search. They type keywords into a search engine (e.g., Google).	The user states their need. They ask a question in natural language or make a request to a conversational agent (e.g., ChatGPT, Gemini, Claude).
2. Discovery & Exploration	Many results and a fragmented journey. The search engine displays a results page with links to e-commerce sites (organic or paid). The user has to click a link, leave the search engine and navigate to the merchant’s site.	A selection displayed right in the conversational agent’s interface. The AI agent queries and aggregates catalogs from several merchants in real time. It then displays a personalized selection of relevant products (images, prices, aggregated reviews) directly in the conversation with the user.
3. Refinement & Decision	Manual action by the user on the site. On the merchant’s site, the user browses the catalog, applies filters, reads product pages and compares reviews (often a time-consuming process). The user then adds the product to the cart.	Ongoing dialogue and AI assistance. The user can talk with the agent to narrow things down: “Show me only the red models” or “Can it be delivered tomorrow?” The agent answers and updates the selection. The user picks a product in the chat window.
4. Order & Payment	“Checkout” process on the merchant’s site. The user has to create an account or log in, enter shipping and payment details, then confirm the purchase on the merchant’s site. Here again, the process takes time.	An integrated, simplified transaction (“instant checkout”). The user confirms the transaction without leaving the conversational agent (using a saved payment method, Apple Pay, Google Pay, etc.). The agent orchestrates the purchase.
5. Completion	Confirmation on the merchant’s site. The user receives a confirmation email from the merchant. The merchant handles payment collection and logistics.	Confirmation in the agent’s interface. The agent passes the order and payment on to the merchant via API. The merchant remains responsible for logistics (order preparation and shipping). The user receives a confirmation from both the agent and the merchant.

Process step

Traditional process (yesterday)

Agentic process (today)

1. Intent & Search

The user starts the search.

They type keywords into a search engine (e.g., Google).

The user states their need.

They ask a question in natural language or make a request to a conversational agent (e.g., ChatGPT, Gemini, Claude).

2. Discovery & Exploration

Many results and a fragmented journey.

The search engine displays a results page with links to e-commerce sites (organic or paid). The user has to click a link, leave the search engine and navigate to the merchant’s site.

A selection displayed right in the conversational agent’s interface.

The AI agent queries and aggregates catalogs from several merchants in real time. It then displays a personalized selection of relevant products (images, prices, aggregated reviews) directly in the conversation with the user.

3. Refinement & Decision

Manual action by the user on the site.

On the merchant’s site, the user browses the catalog, applies filters, reads product pages and compares reviews (often a time-consuming process). The user then adds the product to the cart.

Ongoing dialogue and AI assistance.

The user can talk with the agent to narrow things down: “Show me only the red models” or “Can it be delivered tomorrow?” The agent answers and updates the selection. The user picks a product in the chat window.

4. Order & Payment

“Checkout” process on the merchant’s site.

The user has to create an account or log in, enter shipping and payment details, then confirm the purchase on the merchant’s site. Here again, the process takes time.

An integrated, simplified transaction (“instant checkout”).

The user confirms the transaction without leaving the conversational agent (using a saved payment method, Apple Pay, Google Pay, etc.). The agent orchestrates the purchase.

5. Completion

Confirmation on the merchant’s site.

The user receives a confirmation email from the merchant. The merchant handles payment collection and logistics.

Confirmation in the agent’s interface.

The agent passes the order and payment on to the merchant via API. The merchant remains responsible for logistics (order preparation and shipping). The user receives a confirmation from both the agent and the merchant.

In practical terms, the promise of agentic commerce is to offer, very quickly and through a natural exchange with AI, products sold by every merchant, directly within the conversational agent.

Search becomes easier and the buying journey gets shorter: shoppers are no longer sent to an e-commerce platform that may or may not be efficient or available.

So what are the key takeaways?

“From now on, AI plays a central role in the customer journey. AI is becoming a player in the buying process, and that is not up for debate. So we will have to take its needs and expectations into account, just as we have always done for human visitors.”

Pascal Brunel, Co-founder and President of Synolia

This last point is crucial, and we will come back to it later.

What’s under the hood of agentic commerce?

For this to work, AI needs fast, efficient access to every merchant’s product catalog and, above all, to structured, standardized information it can use instantly to deliver the best possible results in the conversation.

Yet every seller runs its own e-commerce stack (Shopify, Adobe Commerce, Magento, Sylius, OroCommerce and PrestaShop are the most widespread today) and a whole application ecosystem that may include:

a PIM (Akeneo, Quable, Afineo…)
an ERP,
one or more CRM tools (HubSpot is a great fit for e-commerce)
a customer support tool (such as Zendesk, the market leader)
etc.

In short, a multitude of tools, and therefore complex communication between them.

Exchanges between artificial intelligence and merchants’ application ecosystems needed to be standardized. That is why, in fall 2025, OpenAI launched ACP: the Agentic Commerce Protocol.

What is the Agentic Commerce Protocol?

It is nothing more and nothing less than a standardized exchange protocol that lets an AI agent communicate in a structured way with a merchant’s systems (its catalog, its payment system) to complete an end-to-end transaction.

ACP serves 3 functions:

Standardizing product discovery and presentation

ACP makes products readable by AI: it defines how merchants should structure and present their product data (inventory, prices, images, variants, availability) so the AI agent can understand, compare and display them directly in the conversation.
It enables aggregation: it gives the agent the ability to query several merchants’ catalogs at once and offer the user the best selection, much like a price comparison site, but in a conversational way.

Orchestrating checkout and orders

Order flow management: ACP defines the steps to create a cart or a purchase intent, check shipping costs, calculate taxes and start the ordering process, all through structured API calls between the agent and the merchant.
The merchant stays the Merchant of Record: crucially, it lets the merchant remain the official seller (Merchant of Record). The agent is only the intermediary that executes the purchase: the merchant handles payment collection, customer relations (after-sales service) and logistics (shipping and returns).

Securing payment (Instant Checkout)

Delegated, secure payment: ACP often relies on secure payment tokens such as the Shared Payment Tokens developed by Stripe, a major ACP partner. These tokens let the agent pass the payment information the user has saved (in the agent or with a payment provider) on to the merchant without ever exposing the actual card numbers.
Authentication and authorization: the protocol includes mechanisms to make sure the user has actually authorized the agent to make the purchase, and to identify the agent reliably (the Know Your Agent or KYA concept).

“ACP is the technical stack that turns a human-AI conversation into a real transaction between the two agents. It is an open source building block, so digital players will adopt and use it on a massive scale. ACP is set to become a standard that no one can afford to ignore.”

Pascal Brunel, Co-founder and President of Synolia

Agentic commerce: what is at stake for online merchants?

As always, merchants need to adapt to innovation, to these new practices (AI and agentic commerce) and to these new protocols (ACP) to maintain, protect and, if possible, grow their market share.

Merchants face several challenges, and they are both immediate and highly strategic:

The buying journey is changing: the e-commerce site is no longer the only entry point before the transaction.

This does not mean the e-commerce site is going away, for two reasons:

First: the e-commerce site will become the merchant’s transaction hub. It will process its own transactions, but also aggregate transactions coming from other sales channels. Merchants already did this for marketplaces; now they must do it for AI agents too.

Second, the e-commerce site must also convey the brand’s information and values (somewhat less true for a general retailer). It can no longer be a mere transactional vehicle: it has to be a flagship that safeguards the brand’s positioning and carries authority, which also helps the brand rank well in AI engines.

“That is why, in any merchant's growth strategy today, it is essential to see the e-commerce platform as a control tower, a hub for information AND for selling its products. The entire development strategy of the e-commerce platform must revolve around these challenges.”

Pascal Brunel, Co-founder and President of Synolia

The technical challenge is real and needs to be addressed now

Standardized exchange protocols are the kickoff of a deep transformation. ACP must be built into your project right away: your data needs to be exposed through this protocol so you can quickly earn a place in the answers AI agents give to shoppers’ requests.

Just as ranking on Google was every merchant’s priority in the 2000s, ranking in AI is the challenge of today.

Your site must carry authority, be open and easy for agents to query, and let you aggregate direct orders, marketplace orders and orders placed through AI agents.

Get your e-commerce ready for the age of agentic commerce with Synolia

To put this strategy into practice, you need a partner who can address all of these challenges.

Can Synolia help with your e-commerce transformation project and your agentic commerce integration?

At Synolia, we have been preparing for this shift for several years. Our expertise covers every dimension this transformation requires: e-commerce, CRM, BI, PIM/DAM and more.

“AI should now be seen as a “user” like any other, one that will consume your data. The goal is to serve it product information in the right format (ACP), quickly and efficiently. AI does not have the same expectations as a human shopper (it hardly cares about the front end), but both are the players you need to convince to drive the transaction.”

Pascal Brunel, Co-founder and President of Synolia

And because these topics are highly strategic, we work closely with search specialists who are adapting their positioning strategies to the new challenges of AI.
In practice, this means your SEO agency needs to work hand in hand with us to go beyond classic Google campaigns toward a finer, complementary approach: building brand authority and product authority in AI engines.

In France, agencies such as Empirik, Metrixx and Agence WAM are already leading the way on this topic.

Conclusion

If we summed it all up in a table (to forward to your e-commerce or digital leadership, or straight to your CEO!):

Aspect	Traditional process	Agentic process (agentic commerce)
Main point of interaction	E-commerce site (destination)	Conversational agent (starting point and endpoint)
Journey length	Long (clicks, browsing, filling out forms)	Short (direct dialogue, built-in purchase)
Type of query	Keywords (precision required)	Natural language (expressing a need)
Role of AI	Negligible, apart from suggestions on the e-commerce site.	Central and autonomous (search, comparison, purchase)
Challenge for the merchant	Ranking well on the search engine (SEO).	Making your catalog “readable” by AI (GEO, Generative Engine Optimization) and accepting payment through the agent. Building your brand and product authority. Turning your e-commerce platform into a transaction hub.

It is time to seize the opportunities AI offers. Every company has to adapt, starting now, or risk losing big. Conversely, those that take the subject seriously and surround themselves with a knowledgeable partner will be the winners. Shall we talk now, or wait and watch the train go by?