Dashboard to decision-maker: retail’s big shift to agentic AI
ft.com · 25 June 2026
Zalando, an international online retailer based in Berlin, is developing its existing conversational interface into a lifestyle agent for consumer interactions and has recently endorsed Google's Universal Commerce Protocol to be in the first wave when Google brings Agentic Commerce to Europe. Nvidia's 2026 survey found that 89 per cent of retail and consumer packaged goods companies that actively deploy AI said it is boosting annual revenue, while 95 per cent report lower annual costs. The article predicts that by 2028, 60 per cent of brands will use agentic AI to interact directly with individual customers. Nvidia's research found that retail ranks second only to telecommunications in agentic AI adoption, with 47 per cent of respondents using or assessing autonomous AI agents. According to a 2026 Sopra Steria report, 55 per cent of European consumers are aware of agentic commerce; acceptance varies, with 45 per cent willing to delegate electrical and high-tech goods purchases to an AI agent, and only 16 per cent trusting an agent to buy health or medication.
This is the observable fact the cited source supports, stated without interpretation.
Tests HBI mechanism (autonomous systems acting on behalf of users) and the boundary condition of uneven trust by domain. Zalando's interface shift is observable change in activity structure toward orchestration rather than pure human interaction. Revenue and cost claims, though vendor-reported, represent a testable performance assertion under autonomy. The domain-specific variance in consumer trust establishes a real limit on uniform HBI expansion.
This represents early commercial deployment of HBI capability (autonomous systems acting on customer behalf) in retail. The adoption figure of 47 per cent in retail and Zalando's stated transition from adviser to agent-based interface are observable expressions of HBI implementation. The revenue boost and cost reduction claims rest on vendor survey data rather than independent measurement. The variance in consumer trust by category (45 per cent for electrical goods against 16 per cent for health) suggests that HBI adoption is uneven across transaction types, indicating a real constraint on uniform autonomous interaction. The article illustrates that genuine agentic capability differs from labelled-but-traditional RPA, distinguishing true agency from marketing language. Zalando's move represents a shift from HBI (human interaction mediated by AI) toward orchestration of customer choice through agent interfaces, not yet toward value concentration or margin structure change.
This represents early commercial deployment of HBI capability (autonomous systems acting on customer behalf) in retail. The adoption figure of 47 per cent in retail and Zalando's stated transition from adviser to agent-based interface are observable expressions of HBI implementation. The revenue boost and cost reduction claims rest on vendor survey data rather than independent measurement. The variance in consumer trust by category (45 per cent for electrical goods against 16 per cent for health) suggests that HBI adoption is uneven across transaction types, indicating a real constraint on uniform autonomous interaction. The article illustrates that genuine agentic capability differs from labelled-but-traditional RPA, distinguishing true agency from marketing language. Zalando's move represents a shift from HBI (human interaction mediated by AI) toward orchestration of customer choice through agent interfaces, not yet toward value concentration or margin structure change.
This is the framework's reading of the fact above. The source does not endorse it and is not cited as doing so.
The observation establishes that genuine agentic deployment is beginning in retail and that consumer trust varies by domain. The Nvidia survey of revenue boost and cost reduction is current performance claim by vendors and their customers, not yet a test of commercial durability under autonomous operation. Zalando's interface redesign is observable now; whether this structure withstands margin pressure or value capture changes remains longitudinal over the next 1 to 2 years. The 2028 prediction (60 per cent adoption) cannot be tested until that horizon. The variance in consumer acceptance by transaction type is a real early boundary on HBI expansion.
Current evidence is recorded immediately. The paper's longer empirical horizon determines when persistence becomes a stronger framework-level test.
The revenue and cost claims come from vendors and companies that benefit from AI adoption; no independent audit or regulator measurement substantiates these figures. The high awareness figure (55 per cent) may reflect marketing language use rather than understanding of actual autonomous decision-making. Zalando's agent development is a product roadmap statement; the text does not confirm execution, autonomous performance in live commerce, or any margin or revenue consequence from the deployment so far. Consumer acceptance figures reflect willingness to trust, not actual purchasing behaviour through agents.
Recorded for every observation assessed at strength 4 or 5. A record that cannot state the strongest competing reading of its own evidence is not published.
Where several outlets report the same development, they are recorded as additional sources on this one observation and are not counted as separate evidence.