Agentic commerce uses AI to act on behalf of users or businesses. AI agents for ecommerce can make personalised recommendations, manage inventory, and interact with customers. They aim to enhance user experience and boost operational efficiency.
By now, we are collectively coming to terms with the fact that generative AI will change commerce — and the commerce workforce — as we know it. Merchandisers, developers, fulfilment teams, and marketers already use AI for ecommerce to automate routine tasks and boost productivity. But the tools at our disposal will become increasingly more autonomous and intelligent. Agentic commerce is on the horizon.
Agentic commerce is the next phase of generative AI for online businesses. Unlike bots that can carry out simple tasks when prompted by business users, agents will work side-by-side with your teams. Here's how it works.
According to research performed by Gartner , 33% of enterprises will include agentic AI by 2028, up from less than 1% today. That's because agentic AI has the power to usher in a new era of productivity and solve critical business challenges at scale. Rather than past eras of AI that required painstaking prompts and continual fine-tuning from human counterparts, agentic commerce tools have the ability to autonomously pursue goals, make decisions, and dynamically adapt to changing conditions without human intervention. Commerce agents can anticipate needs, make sense of large data sets, and act on insights. In the future of commerce, agents and humans will work side-by-side to create hyper-personalised, high-converting shopping experiences.
Today, we're already used to ‘predictive AI’ – which analyses data to provide recommendations, forecasts and insights–and ‘generative AI,' which learns from data and uses patterns to seamlessly generate text, images, music and code. Agents go far beyond this. They can perform tasks independently, make decisions and even negotiate with other agents on our behalf. And these new AI agents are easy to build and deploy, unlocking massive capacity.
Marc BenioffCEO, Salesforce
The major difference between bots and ecommerce agents is that agents are proactive, whereas bots are simply reactive. Agents can perform higher-order planning, reasoning, and orchestration without needing very much human handholding. In contrast to copilots and chatbots that rely on human requests and struggle with complex or multi-step tasks, ecommerce agents offer a new level of sophistication. They operate autonomously, retrieving the right business data on demand, building action plans for any task, and executing these plans without requiring human intervention.
Consumer-facing agents also take chatbots and other AI shopping interactions to the next level. Unlike traditional chatbots, agents can reason, learn, and adapt. Once agents are integrated with your different sales channels, they can access catalogue, product, customer preference, and behavioural data to deliver personalised, natural sales conversations.
Like a self-driving car, commerce agents use real-time data to adapt to changing conditions and operate independently within customised guardrails. This ensures that every customer interaction is informed, relevant, and valuable. And when desired, ecommerce agents can seamlessly hand cases off to human employees with a summary of the interaction, an overview of the customer's details, and recommendations for what to do next.
Ecommerce agents have unique attributes that enable them to follow through on a task and deliver results that will make an impact on your business. These attributes are:
Role (What job should they do)
Describe, in natural language, the job you want an agent to do. For example, if you notice an item is underselling, you can task an ecommerce agent to create a new merchandising or promotion strategy. Then, the agent will search for semantically related resources in your business metadata. Based on contextual awareness of how your business works, the agent will autosuggest topics and actions for completing the project or task.
Data (What knowledge can they access)
An agent is only as good as the data it can access. Trusted data is the information the agent needs to carry out its role. If you tasked an agent with creating a new merchandising strategy for an underperforming item, the agent might need access to sales data, customer reviews, inventory details, and more. Depending on the task you give an agent, data could also include in-store data, PIM, ratings and reviews, company knowledge articles, CRM data, and more. This means you will need to consider privacy and security as you enable agents to ingest business and customer data.
Actions (What capabilities do they have)
Actions are the predefined tasks an agent can execute to do its job based on a trigger or instruction. For example, it could run a flow, prompt template, or Apex. For example, a merchant can deploy an agent to create a promotion that liquidates excess inventory, targeted to a particular customer group. For many merchants, creating a promotion is a very manual task — and it often leaves money on the table. Agents can intelligently set promotions that meet your business goals and drive customer lifetime value.
Guardrails (What shouldn't they do)
When you create an agent, you know exactly what you want them to do. But you also need to think about guardrails and define, precisely, what you don't want them to do. These can be natural-language instructions to escalate to a human, or could come from built-in security features, such as Salesforce's Einstein Trust Layer.
Channel (Where do they work)
Channels are the applications where agents can do work. This can be your website, messaging app - like WhatsApp, CRM, mobile app, Slack, and more.
Consider all the information an ecommerce merchandising or marketing team would need to uncover to get started: Sales performance for specific products and categories. Marketing engagement performance to determine the success of previous campaigns. Customer data like order history, preferences, and product searches. Gathering this information could take days — sometimes even weeks.
With agents, it can be done in an instant. A merchandiser could simply build an agent to uncover these metrics and datapoints on a routine basis, and suggest personalised promotions for specific customer segments. For example, an apparel merchandiser might build an agent using this prompt: “Suggest weekly promotions to help us sell the three lowest-performing items in each category across our website." Ultimately, agents help you boost sales and simplify marketing with auto-generated, personalised promotions based on real-time business insights.
Boost conversions with AI-powered data analysis. Merchandisers gain insight into shopper behaviour and top-selling products, discovering what shoppers often buy together. They can use AI to take action to improve store performance, grow customer lifetime value and achieve business goals.
Quickly and reliably create copy for product listings based on a deep understanding of your inventory. Agents can also update and refresh descriptions based on customer reviews of specific products, helping you proactively address questions and concerns before customers make a purchase. For example, agents can include unique details such as, “A majority of customers say this shirt runs small. If you're in-between sizes, it's best to size up!”
You can think of shopper agents as a digital concierge, providing personalised assistance by guiding product searches and offering tailored assistance on ecommerce sites, chat, or messaging apps like WhatsApp. Here's how they work.
Product discovery: Customers can find products and make purchases faster using natural language interactions wherever they want to shop (on digital storefronts, in messaging apps, or in retail apps). For example, instead of filtering by category and comparing products, a shopper can simply prompt an agent to find what they need: “I'm going camping in the northwest. Can you suggest gear and supplies I'll need? I want to spend less than $500, total." This provides a more streamlined shopping experience for your customers. The benefit for retailers? Shorter time to browse and improved add-to-cart rates.
Product recommendations: Shopper agents can analyse customer data to offer personalised recommendations to enhance satisfaction and loyalty. For example: if a shopper is looking for a product recommendation for a new pair of running shoes, a product recommendation agent action may suggest certain items depending on commerce data, inventory, and shopper preference like shoe size, race terrain, and more.
Order actions: Shopper agents can also track orders and provide seamless post-purchase support, handling common tasks like “Where is My Order (WISMO)" requests, reducing the need for customer service interactions. It can also handle returns and order modifications. The result? Lower return rates, higher customer satisfaction ratings, and lower cost of service cases.
Scalable. With an increasingly competitive marketplace comes the added importance of outpacing competitors with innovation and growth strategies. Agents are the answer to ecommerce businesses’ scalability dreams. Imagine boosting productivity and making more informed decisions in collaboration with agents who know your business as well as you do. Now, imagine those agents building more agents.
Minimised context switching. It's all as simple as asking for what you want. Enable agents and begin sending prompt requests from one centralised chat window next to a pane that displays what is happening under the hood.
Ecommerce personalisation on demand. Tailoring experiences for your customers just got easier. Rooted in your data, you can trust that when shoppers ask for specific products, accommodations, and service, agents will carry out those actions reliably and adaptively.
Faster time-to-market. With agentic commerce, it's more efficient to get your products in front of prospective customers. As an added bonus, you can devote more focus to more complex, human tasks like research and development, business development, and product innovation.
For many organisations, developing an agentic commerce strategy can feel daunting. In fact, 67% of organisations report that their business lacks a defined AI strategy. And that's understandable: You're navigating uncharted territory and discovering how to use new technology in a way that's meaningful for your business. But right now? Take a deep breath. We're in the dawn of a new era and while it feels overwhelming, we're here to remind you that the challenges businesses you face today are actually familiar ones. But ecommerce agents will help you solve them increasingly quickly, with better accuracy, and at a much larger scale. Commerce Cloud comes with AI agents out-of-the-box, so you can build the ones you need to grow your business the ways that work best for you.
Agentic commerce refers to the use of autonomous AI agents that can act on behalf of customers or businesses to perform complex tasks, such as finding products, negotiating, or managing purchases.
Consumers benefit from highly personalised, proactive assistance, time savings, optimised purchase decisions, and a frictionless shopping experience with minimal effort.
Businesses can use agentic AI to automate customer service, personalise product discovery, manage supply chain logistics, optimise pricing, and enhance fraud detection.
It relies on advanced AI, machine learning, natural language processing, reinforcement learning, and sophisticated decision-making algorithms.
Ethical concerns include data privacy, transparency of AI actions, potential for bias in autonomous decisions, and accountability for agent behaviours.
Agentic commerce is expected to revolutionise online shopping, moving towards more intelligent, proactive, and individualised digital interactions that anticipate user needs.
