The "disruption narrative" around AI agents is once again sweeping through Wall Street, triggering a sharp selloff in internet marketplace platforms. Over the past two trading days, the median decline in Bernstein's basket of related stocks reached 7%, with valuation multiples for companies like DoorDash, Instacart, and Uber coming under renewed scrutiny. However, Bernstein's latest research report clearly states that AI agent commerce remains in its very early stages, consumer behavior shifts are far from materializing, and there is a significant disconnect between market panic and actual fundamentals—creating a buying opportunity for investors.

The catalyst for this round of panic was a series of consumer-facing AI agent product launches. Meta launched its personal AI agent, Muse, on September 8, running on its proprietary Muse Spark model, accessible via a standalone app, web, or WhatsApp. It can send emails, book travel, fill out forms, and complete shopping payments on behalf of users. Muse integrates with Stripe's Link agentic wallet, enabling direct checkout across over one million merchants and using one-time virtual cards to protect users' real payment information. Meta's Chief AI Officer Alexandr Wang told the media that the company views commerce as a potential future revenue stream. Meanwhile, Silicon Valley startup Instinct represents another path. This AI personal assistant, created by 23-year-old founder Noah Shinn, lets users connect their WhatsApp, iMessage, or email accounts with one click, then delegate tasks via text or phone calls—replying to emails, managing calendars, booking flights, comparing insurance prices, and even canceling subscriptions. Instinct opened invite-only beta testing in February and completed a $250 million Series B round on August 26, with its valuation skyrocketing to $2.5 billion—quintupling in just three weeks. In parallel, OpenAI released GPT-6 Astra, which scored 59.3% on the Agents' Last Exam benchmark, capable of operating desktop software, filling forms, and running tests. This led co-founder Greg Brockman to call it "the beginning of the AGI era." Additionally, the continued rise of OpenClaw, Claude Cowork's Dispatch feature, xAI's Grok Bot, and Apple's plan to rebuild Siri with AI capabilities together complete the "always-on" consumer agent narrative mosaic.

Bridging the gap between panic and reality

Bernstein's core assessment is that most of the current discussion remains theoretical speculation. The report notes that despite the widespread adoption of applications like ChatGPT and Gemini, referral traffic from AI chatbots to top e-commerce platforms still accounts for less than 1% of total traffic. Similarweb data shows that in August 2026, Wayfair's GenAI referral traffic share was 1.4%, Walmart's was 1.2%, while DoorDash and Instacart were only 0.4% and 0.3%, respectively. Consumer trust presents another formidable barrier. Checkout.com's "Agentic Commerce 2026" report shows that 33% of consumers expect at least 10% of their purchases to be AI-driven within the next year, but 25% say they will never delegate purchasing decisions to AI, and 27% distrust any institution operating an AI shopping agent. The average single transaction consumers would allow an AI agent to complete without additional approval is only £177, below the £200 merchants expect. By category, consumers are most willing to delegate grocery shopping (41%) and household goods (31%) to AI agents, while financial services only reaches 15%. Notably, 57% of consumers say they would switch brands if an AI agent identified a more cost-effective alternative—suggesting agentic commerce could reshape brand loyalty dynamics.

Traditional platforms' backend fulfillment moat and vertical differentiation

Bernstein's report constructs a framework for evaluating marketplace platform moats, covering six dimensions: supply uniqueness, purchase decision complexity, transaction complexity, competitive positioning, customer loyalty, and excess take rates. Based on this, the degree of impact from AI agents varies significantly across verticals. Ride-hailing and food delivery are considered the most resilient. Uber's liquidity advantage in supply, the breadth of DoorDash and Uber Eats' restaurant delivery options, and Instacart's real-time logistics in groceries all constitute structural barriers that AI agents cannot easily bypass. Subscription-based user stickiness further strengthens defensive capabilities. Traditional e-commerce faces a more complex situation, being relatively more vulnerable to AI price-comparison convenience. Bernstein notes that the highly fragmented retail market paradoxically creates space for AI-assisted price comparison and shopping. However, asset-light platforms like eBay and Etsy need monitoring, as product discovery is central to their value proposition. Wayfair's home furnishing category involves highly visual and emotional SKUs, where logistics capabilities create competitive barriers, but market structure remains the primary concern.

What investors should watch

Bernstein firmly maintains "outperform" ratings on DoorDash (DASH.US), Instacart (CART.US), Uber (UBER.US), and Wayfair (W.US), while assigning "market perform" ratings to eBay (EBAY.US), Etsy (ESTY.US), Lyft (LYFT.US), and Zillow (ZG.US). The report emphasizes that the next 12 months will be a critical observation window—whether consumers truly transition from "learned behaviors" to agent-driven shopping patterns, and whether market platforms can convert AI capabilities into quantifiable revenue growth. The team led by Bernstein analyst Nikhil Devnani draws an analogy to the long-running debate over Uber and autonomous vehicles: market sentiment has swung excessively negative, while the fundamental changes with actual impact still require time to validate. For patient investors, the value created by panic selling may be more worthy of attention than the AI agents themselves.