The Maze: Central Europe is not rejecting AI shopping assistants. It is withholding judgment. A 56% majority is neutral or undecided, while the positive and negative camps sit almost level at 22% and 21%. That makes trust the commercial bottleneck. Retailers do not need to convert an entrenched hostile market. They need to show the undecided middle that an assistant can reduce shopping effort without quietly taking control of the purchase.
The middle is larger than both edges combined. The exact sentiment split is 56% neutral or ambiguous, 22% positive and 21% negative. On the printed figures, the undecided bloc is 13 percentage points larger than the positive and negative camps combined. More important, the two edges are separated by only one point. This is not a polarized market with a dominant pro-AI constituency. It is a market waiting for evidence. Product quality, transparent recommendations and a visible human override can still move the center.
The detailed answers show hesitation, not apathy. The six-step response scale runs from 6% “I want this” and 16% “cool” through 31% “it’s okay,” 25% “so-so,” 10% “weak” and 11% “makes no sense.” That ladder matters. The largest groups are mild, provisional judgments rather than hard rejection. A retailer can therefore treat trust as a sequence: prove recommendation quality, show why products were selected, ask before acting, then earn permission for more automation.
Consumers are more open where comparison is the work. Willingness reaches 41% for adult clothing, 39% for electronics and household appliances, and 35% for groceries. Home and interior products follow at 33%, health and beauty at 32%, and sports at 31%. The common thread is decision effort. Electronics require specification and price comparison. Clothing involves preference and fit. Grocery assistance can optimize health, quality or convenience. AI is easier to accept when it helps structure a choice than when it replaces the choice.
The regional product is still a research assistant. ECDB places most Central European markets at the Research stage: agents search, compare and recommend, while mainstream checkout authority remains limited. That explains why comfort is highest before the transaction. Research preserves the final check-in. Buying removes it. The trust gap will not close through warmer copy alone. It needs explicit merchant identity, current price and availability, explainable rankings, spending limits, confirmation steps, simple reversals and clear liability when the agent gets it wrong.
Why it matters: Agentic commerce will not be won by the assistant that can technically do the most. It will be won by the system that makes delegation feel controlled. Central Europe’s neutral majority is commercially valuable because its opinion is still movable. Retailers should design an autonomy ladder: start with comparison, progress to saved preferences and cart building, then offer checkout only after consent is earned. The moat is not maximum automation. It is trustworthy escalation.
