Agentic commerce is a model of commerce in which AI agents understand a commercial goal, reason across customer and product data, decide what to do next, and take permitted actions to achieve it. For shoppers, that can mean finding and buying products. For brands and retailers, it means coordinating discovery, personalisation and merchandising through specialised agents.
What does agentic commerce actually mean?
Most definitions start with the consumer. A shopper tells an AI assistant what they need; the agent searches available products, evaluates options, recommends a match and, with the right permissions, can complete the transaction. IBM describes agentic commerce as AI agents acting for consumers or businesses to research, negotiate and complete purchases, often without direct human intervention. Stripe frames it similarly — agents finding, comparing and purchasing on a customer's behalf, increasingly approaching merchants as proxies rather than humans browsing directly. Shopify describes shoppers discovering, comparing and purchasing products within a single conversation.
That's correct — but for retailers, it's only half the picture. There are two sides to agentic commerce. On the buy side, agents act for customers: interpreting needs, researching, comparing, transacting. On the sell side, agents act for brands and retailers: understanding the catalogue, customer, inventory and commercial context to determine the most relevant experience or action.
A customer's agent might arrive asking for a fragrance gift under £80 for someone who dislikes sweet scents. A retailer's commerce system needs to understand that intent, identify appropriate products, account for availability and commercial rules, explain the recommendation, and potentially assemble the basket. That's two intelligent systems participating in a commercial decision — not simply search.
Agents act for customers — interpreting needs, researching, comparing alternatives, and increasingly transacting.
Agents act for brands and retailers — understanding catalogue, customer and commercial context to determine the most relevant experience or action.
Agentic commerce vs ecommerce
Traditional ecommerce is built around navigation. The shopper does most of the reasoning — translating a need into search terms, deciding which filters matter, interpreting ingredients or specs, comparing alternatives. The site provides information; the customer assembles the answer.
Agentic commerce lets the customer start with the outcome: “My skin has become dry and reactive — I want a simple routine under £100,” or “I need something for a September wedding in Italy, nothing black, nothing sleeveless, and I want to wear it again.” The system interprets the requirement, understands the relevant catalogue, evaluates constraints and determines what should happen next.
How is this different from generative AI?
Related, not interchangeable. A generative model produces an output — ask it to explain niacinamide and it gives an answer. An agent works toward an objective. Tasked with helping someone build a routine for pigmentation and sensitive skin, it may need to:
- Understand concerns and constraints, and determine if more information is needed
- Inspect relevant products and ingredients, and exclude unsuitable combinations
- Factor in previous purchases, assemble a routine, and explain the selections
- Check availability and add items to a basket
The language model supplies some of the reasoning; the agent is the system performing the task. Agentic commerce isn't a chatbot bolted onto an ecommerce site.
Agentic commerce vs chatbots, personalisation and automation
These can look agentic without being agentic. A chatbot primarily responds. A recommendation engine predicts. A rules engine executes predetermined logic. A workflow follows a fixed sequence. An agent reasons about the situation and determines the next action within its boundaries — decisions that used to require thousands of hand-coded rules can instead be handled dynamically with context, reasoning, tools and permissions.
Why beauty and fashion are particularly interesting
Their catalogues combine enormous choice with highly personal decisions. A moisturiser's suitability depends on skin type, concerns, sensitivities, ingredients, climate, existing routine, price and preference. A fashion purchase involves fit, proportion, occasion, weather, style, existing wardrobe and budget.
Traditional ecommerce compresses this into filters — brand, price, size, colour, category — but that's rarely how customers think. They think: “What will work for me?” Agentic systems can reason over several pieces of context at once instead of asking the customer to translate a human need into database filters.
[Christina to source]
What does an agentic commerce system look like?
Intent understanding
“Something for redness” and “my skin gets red after retinol” share a word but describe different problems.
Product intelligence
Reliable structured data on ingredients, claims, compatibility, sizing, fit and styling relationships.
Customer context
With consent, preferences, purchase history, loyalty status and diagnostic information move the experience from generic to contextual.
Reasoning
Evaluating available information against the customer's objective and the retailer's rules.
Tools and actions
Querying inventory, running a diagnostic, virtual try-on, building a basket, applying loyalty benefits.
Permissions and guardrails
What an agent can access, what it can do, and when a human takes over.
One agent or many? — Aria orchestrating specialist agents
Some tasks suit a single agent; complex environments benefit from specialists coordinated by an orchestrator. In ARview's architecture, that orchestrator is Aria, coordinating eight specialist agents. The customer doesn't experience eight disconnected tools; they experience the brand helping them get something done, with Aria deciding which capability the moment calls for.
Worked examples
“My skin feels dry and dull, but vitamin C sometimes irritates it. I want a three-product morning routine under £120.”
[Christina to source]
An agentic experience understands the concern, sensitivity and budget; assesses skin characteristics if diagnostics are available; searches the actual catalogue; reasons across ingredients and compatibility; excludes conflicting products; builds a three-step routine; explains the selections; checks stock and pricing; adds it to the basket.
“I need an outfit for a work dinner next week. Smart but not corporate, under £250, and I don't wear heels.”
[Christina to source]
An agent treats the request as a goal — reasoning across occasion, style, budget, sizing and availability, constructing complete looks, letting the customer refine conversationally, using virtual try-on to reduce uncertainty.
What changes for retailers
The catalogue becomes knowledge, not just pages. Customer context becomes more valuable. Commercial systems need to become actionable — giving an agent the ability to understand why a customer might churn and execute an approved retention action. And the change is organisational: when agents operate across discovery, CRM, loyalty and transactions, the boundaries between systems stop mattering to the customer.
What agentic commerce is not
- ✗ A chatbot added to an ecommerce site
- ✗ A new name for product recommendations
- ✗ Automation wearing an AI interface
- ✗ Unrestricted autonomous decision-making
- ✗ A mandate to rebuild the commerce stack from scratch
The shift from interfaces to outcomes
Ecommerce has spent decades improving interfaces — better search, better filters, faster checkout.
A shopper expresses an objective; agents interpret it, reason over context, coordinate the right capabilities and take permitted action. For beauty and fashion, where the right product has always needed more context than a category page captures, that shift matters more than most.
Frequently asked questions
What is agentic commerce in simple terms?
What is an AI commerce agent?
What's the difference between agentic commerce and ecommerce?
What's the difference between generative AI and agentic AI in retail?
Is a chatbot agentic commerce?
Can agentic commerce work with an existing ecommerce platform?
Why is agentic commerce relevant to beauty?
Why is agentic commerce relevant to fashion?
See agentic commerce built for beauty and fashion
Aria and eight specialist agents, live across skin diagnostics, try-on, routines and more.
Explore ARview's agents