How AI shopping assistants are replacing persuasion with prediction – and why digital commerce faces a warmth deficit

It was less than a century ago that families proudly hoped their children would grow up to be salespeople. Selling meant white-collar jobs rather than factory labor, offered upward mobility, and demanded the charisma and polish that would surely be useful in other areas of life. 

The archetype of the salesman empowered the American worker from itinerant amateur to trained expert, shaping an economy of executives rather than peddlers and canvassers. Now, such prowess resulted in the most powerful economy of the world – and, lest we forget, a history-shaping global technological hub. 

Yet, just as quickly as the salesman emerged, the landscape shifted – the modern consumer no longer encounters a salesperson, but rather a system. 

“Agentic AI systems are reaching an imperative level; to maintain competitive stance, most sales organizations have to embrace some form of agentic AI,” noted University of Mississippi professor Gary Hunter. 

Hunter’s research, published in the Journal of Business Research in January 2026, found that, as the market for autonomous AI agents grows from $7.6 billion in 2025 to over $139 billion USD by 2033, it is the most transformative trend in personal selling and sales management ever. 

Whereas from 1950 to 1980 sales was one of the fastest-growing occupations in the U.S., its popularity declined with the internet boom of the late 1990s and early 2000s. And now, as signaled by EY, AI has been placed at the very heart of the once heart-driven job.

Just last year, the consulting firm found that 67% of people globally use AI as a pillar of their customer experience; LLMs can power personalization at scale while significantly decreasing customer acquisition costs and strengthening retention. Meanwhile, retailers already use AI for pricing, promotions, supply chain decisions, customer service, and supplier management. 

Seemingly, technology has pushed societies beyond the warmth of a greeting once one enters a store, the emphatic gratitude of customers as their shopping assistant hands them the perfect briefcase, and the smiling faces of families who have bought their first van in an auto dealership. 

Deloitte paradoxically stresses that sales is no longer about selling products to customers, but rather building a community hub and offering meaningful interactions; 80% of shopping still happens in physical stores, at a time when 54% of retailers rank the growth of their digital commerce as a top priority for 2026. 

Thus, a contradiction: while forward-looking executives scramble to adopt AI-first strategies, the everyday consumer experience suggests a far less dramatic shift. And, as 49% of sales executives report cost benefits from AI activities, the question of whether the salesman has run out of business is more critical than ever. 

“We are witnessing the death of the digital salesman and the rise of the system. While AI like [Amazon’s] Rufus or [Walmart’s] Sparky excels at predictive logic, they often fail at emotional resonance,” said Martin Lewit, SVP Corporate Development at global technology consulting firm Nisum. 

“We have replaced discovery with automation, creating a warmth deficit where the system remembers your purchases, but not you.” 

The System Takes Hold – Sometimes

Just on February 10, 2026, San-Francisco based AI agent startup Simple AI announced it had raised $14 million in funding, which will be used for the development of its voice agent platform, custom generative AI models, and customer insights for businesses. 

“Building voice agents that outperform human sales teams is a complex task, and I’m immensely proud of what our team has achieved,” said Zach Kamran, co-founder and CTO of the firm. 

What is novel in the U.S. has already been thoroughly documented in China, where AI avatars are used to sell products to Chinese customers through round-the-clock livestreams – and beginning to outperform their human counterparts. PLTFRM, a Shanghai-based firm, has now deployed over 30 AI avatars across selling platforms, creating its virtual salespeople by combining Baidu’s AI video models with DeepSeek’s LLM. 

The surge in AI-supported sales operations – especially in China’s booming livestream commerce industry – signals something deeper: AI outperforms salespeople in speed, personalization at scale, 24/7 availability, and price optimization. 

Another case study by software company Landbase found that AI agents in California are performing faster lead response, higher conversion rates, and scalable outreach – resulting in 70% higher lead conversion and a reduction of costs from 40% to 60%.  

“This is a wakeup call for anyone in sales. AI will increasingly handle qualification, objection handling, personalization and follow ups, which means execution becomes commoditized,” said Mikhail Yerganjiev, business development manager at blockchain security auditor CertiK. 

“What doesn’t get commoditized? Judgement, trust, relationships, domain expertise, knowing why deals close – not just how,” he added. 

“Trust in digital commerce is shifting from ‘people-based’ to ‘process-based’. While this builds confidence in the delivery, it erodes the emotional bond between the brand and the person behind the screen,” Lewit added. 

Salespeople still have a leg up in complex emotional negotiations, high-stakes purchases, and, more critically, trust-building over time. Amazon’s AI assistant Rufus, for instance, was launched in 2024 to make shopping faster and easier through personalized product recommendations and prediction. With its 2025 upgrade, the agent is now “more capable, more conversational and helpful” via upgrades in intelligence, reasoning, and performance. 

Similarly, Walmart’s virtual assistant Sparky, introduced in early 2025, synthesizes reviews, offers occasion-based recommendations and helps customers plan, compare and purchase confidently via its reordering, service booking and accessibility features. So far, it has increased order value by 35%, with at least 27% of its consumers now trusting AI-powered recommendations and alerts.  

How Rufus or Sparky would shine in contexts where the stakes are higher than shopping for a facemask or a lamp remains critical: in enterprise B2B deals worth an average of $50 million; in the slowed-down luxury retail market; or in healthcare sales, operating at the intersection of medicine, value and vulnerability. 

“The trend in e-commerce is hyper-efficiency at the cost of intimacy. Algorithms are great at anticipating needs, but they are currently incapable of anticipating feelings. This creates a digital experience that feels functional but cold,” Lewit stressed. 

The Cost of a Warmth Deficit 

For all of its computational precision, AI thus runs into a deeply human problem: consumers know when warmth is being performed rather than felt. Researchers have identified what might be called an uncanny valley of automated empathy – the unsettling middle ground where a system is responsive enough to feel almost human, but not enough to feel genuinely so. 

While anthropomorphic features like perceived empathetic ability can build user trust in AI agents, the effect reverses when that empathy becomes too expressive; chatbots that apologize through emotional language paradoxically generate more negative evaluations than those that don’t. 

The problem, then, is not that consumers reject AI outright, but rather that they resent being fooled. When a system says “I understand how frustrating that must be,” the consumer doesn’t feel understood; they feel managed. And such a distinction matters commercially. 

“Current AI strategies emphasize productivity for the seller rather than empathy for the buyer. The next frontier isn’t faster delivery; it is ‘humanized automation’, AI that feels accountable and modulated to a user’s emotional state, said Lewit. 

The landscape is far from the apocalyptic “death versus survival”. Now, stakeholders must question what a salesperson’s role actually becomes inside an AI-first organization. Are they relationship managers? Brand ambassadors whose job is emotional theater? Or, conversely, are they being upskilled into something entirely new – part analyst, part consultant, part community builder? 

The warmth deficit is thus not merely a philosophical concern; it has measurable human consequences that are not distributed equally. When a system rises to replace a salesperson, the loss is more acute for those who depended on tangible interactions. From older consumers navigating a world designed for digital natives and low-income shoppers without access to devices or connectivity, to people with disabilities for whom an adapting human presence is a necessity rather than a luxury. 

And, of course, impacts on labor and the economy are as pronounced: the American salesperson was never just an economic actor. They were – and are – a social archetype that embodies self-made success, charisma converted into livelihood. The U.S. Bureau of Labor Statistics projects declining sales employment through 2033, driven largely by e-commerce and automation. 

However, 65% of consumers say they will switch brands if a company does not offer customized experiences – while algorithm-driven personalization produces very different emotional responses than personalization offered by a human. 

Thus the question emerges: how to balance innovation and economic profitability while offsetting the warmth deficit. At the broadest cultural level, commercial interaction in the form of store visits, sales calls, and negotiations were always also social ones. 

The most forward-looking firms are not choosing between AI and human salespeople, but redesigning their relationship. The emerging model positions AI as infrastructural, while returning humans to what they always did best: showing up in moments that matter. 

So, as for the family that hoped for a salesman? Now they most certainly are hoping for the truest version of just that: a child who grows up to be charismatic, driven by human connection, and aided in their quest by emerging technologies. 

Article co-authored by Salomé Beyer Vélez