SALESFORCE

From omnichannel to agentic commerce: How to prepare Salesforce Commerce Cloud for AI shoppers and agents

Author: Daniel Herranz, Salesforce Solution Lead at Inetum
Publication date: November 12, 2026

 

For years, digital commerce teams have worked to remove the boundaries between physical stores, ecommerce, mobile apps, and customer service. Unified catalogs, inventory visibility, click and collect, and cross-channel returns have moved from competitive advantages to baseline requirements for retailers.

 

The next shift is already underway. A growing share of product discovery is happening through AI assistants that can interpret needs, compare alternatives, and recommend purchases. 

People clapping

Digital commerce traffic is changing

Search engines and social platforms remain important, but they no longer account for the entire discovery journey. A consumer can ask an AI assistant to find walking shoes, compare several options below a certain budget, and determine which pair best fits their needs.

This shift is more than conceptual. During Cyber Week 2025, Salesforce reported that AI and agents influenced 20% of the global orders it analyzed through personalized recommendations and conversational customer service. The company attributed $67 billion in sales to those interactions.

That means a growing share of purchase intent is no longer generated exclusively through traditional search and site navigation. AI-assisted product discovery is becoming another path to the transaction.

That creates a new visibility challenge for retailers. When a product catalog has incomplete attributes, ambiguous descriptions, inconsistent categories, or outdated inventory data, the problem goes beyond making the human experience harder. An AI agent may also struggle to interpret, compare, and recommend those products accurately.

Search engine optimization therefore has to coexist with a newer discipline focused on generative engines. The goal is no longer simply to rank in search results. Retailers also need to provide structured, reliable information that allows an AI assistant to consider their brand and products when generating recommendations.

The visitor arriving at a digital storefront is no longer always a person browsing categories. It may also be an AI agent looking for products on that customer’s behalf.

For retailers, the question is no longer just how to deliver an omnichannel experience. It is how to prepare Salesforce Commerce Cloud so customers and AI agents can understand the catalog, receive reliable recommendations, and complete purchases. That means rethinking the experience layer, data quality, and the underlying commerce architecture.

Agentic commerce describes a model in which both people and AI agents can discover products, evaluate alternatives, and complete transactions using shared, up-to-date information.

From searching for words to expressing a need

Most ecommerce experiences still require shoppers to translate what they want into keywords. But customers do not always know the exact product name they need.

They may be looking for “a gift for someone who hikes, for less than $80,” or “two lightweight shirts for work this summer with a total budget of $50.”

An AI shopping agent can interpret that intent, apply budget constraints, check availability, and explain why a particular option might be a good fit. The difference from a traditional chatbot is important: instead of responding from a predefined set of questions and answers, the agent can consult real information about products, inventory, orders, and preferences.

Salesforce has evolved its B2C Commerce offering toward Agentforce Commerce, with guided shopping capabilities designed to support customers from discovery through ordering and post-purchase service within the same conversation. The company has also introduced tools that allow external agents, with authorization, to access current catalog information, images, and shopping carts.

The implication for retailers is straightforward. If customers start delegating more of the discovery and evaluation process to AI, commerce platforms need to make product and transactional information understandable not only to people, but also to the agents acting for them.

A personal shopper connected to the real product catalog

One practical example is a generative personal shopper developed by Inetum for a Spanish menswear retailer. The assistant is integrated into a Salesforce B2C Commerce site and queries the company’s active catalog to handle more complex shopping needs.

The system can reason across variables such as budget, fabric, season, and intended use. It can also analyze an image supplied by the customer, identify the garments that make up a particular look, and find available alternatives. Because it works with real product data and applies filters designed to prevent recommendations for unavailable products, the conversation becomes an extension of the commerce experience rather than a disconnected chatbot interaction.

Its role goes beyond answering questions.

When a customer adds the jacket from a suit, for example, the assistant can suggest the matching pants and account for current promotions. Contextual cross-selling can therefore move beyond products that happen to be broadly related and instead consider the coherence of the outfit and the customer’s purchasing context.

Another use case addresses one of fashion ecommerce’s most significant costs: returns. If a shopper adds the same item in two sizes, the assistant can reference official size charts and use that customer’s order history to identify their usual size, helping them make a decision before checkout.

The objective is not simply to improve conversion. It is also to reduce multi-size orders and protect margins from the cost of reverse logistics.

Building an architecture that can change quickly

Adding AI agents on top of a rigid or fragmented commerce platform can increase technical debt. Artificial intelligence needs access to consistent information, as well as an experience layer that teams can evolve without rebuilding the entire digital commerce environment.

Storefront Next, generally available since June 2026, addresses this challenge with a full-stack architecture built on React, server-side rendering, and client-side navigation. Salesforce says retailers can provision a production-ready storefront site in less than 30 minutes, and the platform includes capabilities for working with AI-assisted development tools.

For retailers, the practical benefit is the ability to reduce the time required to launch and modify experiences. Business teams can manage product pages, listings, and components without depending on development work for every change, while technical teams retain the ability to extend the platform.

For organizations already running an earlier Salesforce B2C Commerce architecture, modernization does not necessarily mean discarding the existing commerce engine. The experience layer can evolve while the underlying engine remains in place.

That distinction matters. The objective should not be to migrate simply because the technology landscape has changed. Retailers need to determine what should be retained, what should be modernized, and which use cases provide enough business value to justify the investment.

How to measure the value of agentic commerce

The success of an agentic commerce strategy should not be measured by the number of conversations customers have with an assistant. The meaningful metrics are still the ones that affect the business.

Conversion can be evaluated by comparing customers exposed to guided shopping with those following a conventional journey. Average order value can show whether contextual recommendations increase basket value. Return rates can indicate whether better guidance around sizing or product compatibility is resolving uncertainty before it turns into a logistics cost.

Retailers can also measure reductions in inquiries reaching the contact center and the time required to launch new digital experiences.

Conversations create another potentially valuable source of information: signals about customer intent. Budget, purchase occasion, preferences, and reasons for rejecting products can be integrated into Salesforce Data Cloud and Marketing Cloud to improve segmentation.

Over time, organizations will also need to watch their presence in responses generated by external AI assistants. How frequently does the brand appear for a relevant shopping intent? And what is the quality of the product information being used to represent it?

These questions expand digital visibility beyond the traditional SEO model.

Data quality and governance remain the starting point

An AI agent cannot fix a poor product catalog on its own. If attributes are incomplete or inventory is out of date, the resulting recommendations will be less useful and can erode customer trust.

That is why catalog and data quality need to be addressed before conversational capabilities are expanded.

Governance is equally important. Retailers need to define what the assistant is allowed to recommend, which promotions it can offer, when it should hand a conversation over to a person, and how its decisions can be audited.

In practice, these rules amount to a codified commercial policy. Establishing them requires participation from business, technology, and legal teams rather than treating agentic commerce as an isolated AI initiative.

Ultimately, organizations need three foundations: reliable data, flexible architecture, and agents that can act on real business processes.

Ecommerce has spent years optimizing a funnel designed for people who search, browse, and click. Now that funnel is starting to include conversations and delegated decisions.

Preparing for that shift does not mean replacing omnichannel commerce. It means building on it for the next stage, making commerce understandable to customers, transactable by AI agents, and relevant when both are deciding what to buy.

Do you want to Learn how Salesforce Commerce Cloud can evolve for agentic commerce, AI shopping agents, better product data, guided buying, and flexible storefronts?