How AI Agents and Human Insight Will Transform the Shopping Experience
E-commerce is a tough business. The margins are often wafer-thin and competition is intense. Consumers have elevated expectations. Success demands a frictionless experience on the front end, along with hyper-efficient operations on the back end.
For as long as anyone can remember, e-commerce businesses have tried to meet their efficiency and user experience (UX) goals by tinkering at the sides. A redesign here, for example, or a new CRM there. These iterative changes helped, but a more transformative revolution is shaping up in 2025, driven by artificial intelligence agents.
It's tough to overstate how significant AI agents will be to online retailers. These AI models will effectively mimic human behavior, combining the latest advancements in large language model (LLM) technology with the ability to process multiple modalities such as image, audio and video. When supported by human validation throughout both the model development and execution processes, AI agents can function like human employees — employees who truly understand your brand. This oversight ensures that the AI's actions are accurate and aligned with the brand’s values and goals.
AI agents will help customers through their buying journeys, providing recommendations and advice in a way that feels personal and brand-specific. And, with real-time access to inventory and fulfillment data, we can expect to see AI agents assume operational roles in e-commerce.
Crucially, AI agents will be able to deliver on a long-term aspiration for online retailers — namely, the ability to replicate the attributes of a real-world sales assistant, but in a way that scales across potentially millions of customers.
We've seen chatbots before. This is something entirely different. AI agents won't be constrained by the number of questions they can understand. They will be able to engage with the customer in a dynamic, warm way that embodies the organization's brand values. By integrating them with other systems, retailers will be able to provide self-service mechanisms, thus reducing UX friction and the demands on human customer support employees.
I'm not the first person to make these observations or predict that 2025 will be the year AI agents start their ascent into the mainstream. Salesforce describes AI agents as part of the "third wave of the AI revolution." Deloitte expects a quarter of all businesses using generative AI to also use AI agents by the end of this year, doubling to half by 2027.
I encourage organizations adopting AI agents to consider their deployment and management throughout their lifecycle carefully. It’s easy — and tempting — to view them as replacements for human staff. However, the most successful adopters will be those who see AI agents as augmentations to their existing workforce, recognizing the vital role of individuals in overseeing these agents throughout their lifecycle.
In the retail context, humans will play a pivotal role in defining the operational scope of AI agents based on real-world objectives and recognized customer needs. For example, when processing customer returns, an AI agent must navigate the return policy accurately. It should be able to assess when to make exceptions for customer satisfaction or detect potentially fraudulent returns. This is evident in how companies like Amazon.com and Walmart leverage AI to enhance customer service while adhering to strict return policies.
AI agents can enhance personalized shopping experiences, but traditional machine-learning models also play a significant role. For example, Stitch Fix utilizes machine learning to tailor clothing recommendations based on individual preferences. While this approach is effective, human oversight remains essential to ensure that suggestions align with brand values and quality standards. In scenarios where AI agents are implemented (e.g., processing returns through a return management system), human staff must still verify the agent's performance to ensure it meets expectations while maintaining the brand's integrity. This collaborative approach not only enhances customer experiences but also helps retailers like Target and The Home Depot optimize their marketing initiatives and operational efficiency.
This AI development and deployment model is known as humans-in-the-loop (HITL), recognizing that the current sophistication of AI often falls short of fully autonomous operation. It requires individuals with domain expertise and an understanding of the brand ethos to serve as quality control and safeguarding mechanisms. These experts ensure that AI performs accurately and aligns with the company’s values and objectives. Understanding this is the difference between failure and success.
E-commerce platforms that approach AI agents with the mindset of solving specific problems and reducing customer friction while understanding that any AI technology requires supervision throughout its life cycle will have the best outcomes. They will be empowered to deliver experiences quickly that are profoundly personalized and laser-focused on successful outcomes. This, in turn, will foster a new level of customer satisfaction and loyalty.
Lisa Avvocato is the vice president of global marketing at Sama, a provider of data annotation solutions that power the AI models of the future.
Lisa Avvocato is a veteran product marketer/moderator specializing in AI and ML technologies. She’s passionate about the interaction of machine learning and digital transformation strategies to reduce inefficiencies and drive sustainability. With over 15 years of experience in Enterprise SaaS technology, she has worked across a diverse set of industries including retail, education, manufacturing, and healthcare. She received her MBA from Hofstra University.
