The Future Of AI In Retail Goes From Personalisation To Prescription With Avatars, Agents, And Automation 

Imagine waking up to a package at your doorstep. An outfit you never picked, yet somehow, it’s exactly your size and your style. After a brief chat the day prior, a series of AI models had analysed your preferences, predicted your needs, and bought it for you. No checkout, no decision making. Just seamless, autonomous commerce.

For all the effort retailers and DTC brands have put into streamlining the fashion and beauty eCommerce experience, from search and product discovery to checkout, the industry has stayed on one side of a firm line: the final choice of what to buy, and how, sits with the consumer. While AI recommendations, virtual try-ons, and predictive analytics have worked to take the friction out of the shopping experience, these tools end at the point of making proactive recommendations and steering shoppers in a particular direction. AI may influence, but it does not decide.

That line is now blurring. OpenAI’s latest announcement (and the company’s first hands-on demo that focused on a set of fashion and retail use cases) signals a shift from AI-powered recommendations to AI-powered decision making. With new agentic features designed to act on behalf of users, AI is poised to move beyond suggesting purchases – it may soon be selecting, purchasing, and managing transactions autonomously. 

For some, this is the ultimate convenience: an AI assistant that takes care of shopping without the need for hassle! For others, it raises concerns about control, choice, and the potential for over-reliance on automated decision making.

This brings us to a critical question: Are we blithely wanding into an era of effortless convenience or prescribed consumption?

On Tuesday, OpenAI released “the first set of building blocks” for developers and enterprises to create AI agents capable of taking multi-step actions. These features, integrated into OpenAI’s API, enable applications to perform web searches, conduct deep research, and even operate digital tools. While this sounds like an incremental step, the implications are vast.

One demo showcased a fashion style agent that scans the web for clothing items based on subjective criteria, compares prices across multiple retailers, and theoretically, adds selections to a cart for checkout, all without direct human input. 

If this is where AI is headed, then shopping may soon shift from an active decision-making process to a passive, behind the scenes automation. The very experience of browsing, searching, and selecting products could give way to a world where AI preemptively makes choices, streamlining retail but fundamentally altering the nature of consumer engagement. 

What makes this shift even more intriguing is the integration of hyper-realistic digital avatars into the experience. If AI agents are making purchasing decisions, why not give them a human-like presence?

AvatarOS, a company that just secured $7 million in seed funding, is developing AI-powered avatars with the personalised traits and lifelike movements. Unlike static chatbots, these avatars are intended to interact in real time, powered by large language models. 

While their current focus does not explicitly include developing avatars to serve as personal stylists or retail assistants, focusing instead on the virtual influencers space, the jump, we feel, wouldn’t require a huge amount of development. And to make things even more realistic, the newest research demo from Sesame (open for anyone to try) is aiming to achieve “voice presence” for digital assistants, a feature The Interline tested and found compelling as a potential breakthrough in conversational interfaces built on top of LLM’s. 

What does all this lead to? An increasingly feasible outcome is an AI agent that understands consumers preferences, provides recommendations, and engages in natural conversation, all behind the veneer of a bespoke digital avatar. 

But this is about more than just making AI feel human, crucially, what do people want to believe is behind that friendly face? As AI-generated personas and influencers grow more sophisticated, the line between authentic and synthetic interactions will blur, raising deeper questions around emotional connection and consumer agency.

And as AI avatars become more lifelike, the boundaries between digital assistants and human persuasion could start to blur, with implications for how we think about the way we represent people digitally. Josie Pearon, Agency Manager at The Bureau, a company focused on representing real world human models in virtual environments and digital applications, shared her thoughts with The Interline.

“We have the opportunity for real people to be celebrated and included in virtual environments, rather than replicated or replaced.“There is a lot of mystique and allure surrounding virtual influencers, models and humanism which ,whilst an attractive area of exploration for brands, risks consumers being exploited via emotions and biases, making people more susceptible to persuasion.” 

And we could soon face a very sci-fi-sounding but nonetheless feasible scenario, where AI-powered personas are guiding consumer choices without that consumer knowing what share of authentic humanity sits behind them. “If consumers are unaware that they’re interacting with AI, it raises transparency issues. We believe that real people and their experiences and interactions hold the ultimate value when it comes to preserving and encouraging brand loyalty,” Pearson adds.

The move towards AI-driven and, potentially, AI-executed retail presents something of a paradox, then: personalisation should enhance self expression, but at what point does it start replacing it? If an AI stylist predicts fashion choices more accurately than the consumer themselves, is it still a reflection of personal style? If AI shopping agents analyse behavioural patterns to predict and execute purchases, does the individual remain an active participant in their own shopping journey? And if preference is inferred rather than explicitly expressed, does personal agency give way to passivity?

Convenience can only stretch so far before it diminishes the shopping experiences altogether. For some consumers, the thrill lies in the hunt, the dopamine hit of discovery, impulse buys, or finding that perfect item. Can AI replicate that emotional experience or does it sterilise it?

From a business perspective, AI shopping agents introduce both opportunity and risk. The evolving role of AI in commerce is not just about efficiency, its’ also about control. Open AI’s models – as made available through its own APIs – enable AI agents the autonomy to browse across multiple retailers, favouring price, availability, or, and this is key, algorithmic preferences over brand loyalty. This means brands will compete not just for human customers, but for AI agents acting on their behalf. 

One counter strategy could be for brands to develop their own AI shopping assistants using the tools that OpenAI and others are making available, ensuring product visibility, maintaining brand storytelling, and fostering a guided decision-making process aligned with its brand values. The future of AI commerce may depend on who controls the algorithm: the brand, the AI provider, or the consumer. 

While brands may still have time to adapt, technology’s prescriptive nature is already well underway. We talk a lot about the inherently opaque nature of some aspects of the fashion industry, but AI decision making presents its own “Black Box Problem”. Right now, OpenAI allows users to halt an agent mid-process and reveal its decision making logic. But for how long?

As OpenAi’s own demo suggests, transparency may soon fade into the background, making AI driven shopping less about conscious decision making and more about automated, unseen transactions. Once we, as consumers, become accustomed to AI making choices on their behalf, will we ever take back control? This is where brands must proactively shape their AI strategy, ensuring whatever shopping agents emerge they remain aligned with consumer expectations and brand integrity. In this new world of AI-powered retail, the greatest competitive advantage may not be price or product, but trust.   

Best from The Interline:

Kicking off this week, Lauren (ALTRA) Kacher on how brands may not be universally rushing to sell digital-only goods right now, but from technology to process, digital and multi-channel “phygital” fashion initiatives could still have a lot to teach fashion about how to approach digital product creation.

In The Interline’s first Tuesday news instalment: beauty is exploring a unique opportunity to sell mechanical innovation directly to consumers, but will that model scale? And in fashion, behind-the-scenes Silicon Valley investment and open commitments from luxury are testing new waters with AI.

Tukatech’s CEO and Founder on why 3D design and digital product creation remain a compelling investment even in challenging economic times, and how the business case for digital transformation is evolving to meet industry demands.

Closing out the week, Interloop’s Mohammad Amir writes on how, while the attention has been focused on brands’ and retailers’ 3D strategies, the world’s biggest suppliers have been making milestone commitments to testing new frontiers of DPC in preparation for industry-wide transformation.