Retail — AI Deployment

Retailers deploy AI across product discovery and personalization, customer-facing shopping assistants, pricing, supply-chain and inventory management, and drone delivery. Several large deployments have been rolled back after live operation revealed reliability shortfalls, and algorithmic pricing has drawn state legislation and legal challenges.

Adoption patterns

Major retailers apply AI to customer-facing discovery and personalization, back-end inventory and supply-chain functions, and last-mile delivery. The following deployers are documented by application:

Shopping assistants and agentic commerce

Retailers and platform providers have built AI shopping interfaces into their products. Amazon's Rufus in-app shopping assistant reached general availability in February 2024. OpenAI introduced "Shopping Research in ChatGPT" on November 24, 2025, positioning ChatGPT as a shopping interface. Analyst coverage of the trend includes "How shopping chatbots might transform retail" (January 29, 2026).

Chatbot-mediated purchasing raises a purchase-agent liability question: if a chatbot buys the wrong item, who is responsible.

On May 27, 2026, AWS packaged the architecture, starter code, and learnings behind Alexa for Shopping into a service that lets retailers launch their own AI shopping assistants "in as little as 60 days." Tapestry-owned Kate Spade was the first announced customer, deploying a gifting assistant. AWS described the offering as an alternative to retailers ceding their storefront to OpenAI, Google, or Perplexity shopping agents, framing it as "first-party agentic commerce" infrastructure that gives retailers a defensive option against AI-agent disintermediation of the customer relationship. See Amazon and Agentic AI. (Source: cnbc.com)

DoorDash launched a conversational AI assistant on June 11, 2026 that lets customers order food and book reservations using prompts and photos, extending agentic commerce into food delivery (Source: techcrunch.com).

A study published July 30, 2026 by Columbia Law School's Center for Law and the Economy examined what shopping assistants detect but do not act on. Researchers found that Amazon's Alexa for Shopping and Walmart's Sparky can identify mismatches between an explicit "Made in USA" claim and contradictory country-of-origin information elsewhere in the same listing, but that neither retailer uses the capability to flag or remove the listings. Asked why, Alexa told researchers that "the harm to U.S.-made brands is real and documented, but until that harm creates a financial, regulatory, or reputational cost for Amazon specifically, it remains easier to do nothing"; Sparky cited FTC practice of enforcing against manufacturers rather than retailers, adding "That's a business calculation, not a legal justification" (Source: reuters.com). FTC rules permit the label only where a product is "all or virtually all" made domestically, and the agency urged both retailers to police third-party sellers' claims in 2025. An Amazon spokesperson said country-of-origin information is displayed on product detail pages where available; Walmart did not comment. The study is the center's first, following Lina Khan's return to the university (Source: qz.com).

Six banks, ASB Bank, Bank of America, Capital One, Commonwealth Bank of Australia, ING Group and NatWest Group, published a principles paper, "Building Trust in Agentic Commerce," on September 22, 2026. It states that consumers "are concerned that AI agents may buy the wrong thing or spend too much – or even worse, lose their money to scams and fraud," that merchants expect disputes and chargebacks to rise for reasons outside their control, and that agentic commerce carries "potential for higher rates of scams, fraud and disputes." The paper proposes five voluntary, non-binding principles (transparency, safety, privacy and data, choice, and interoperability) while describing the banks as "excited by the promise of agentic commerce"; Gizmodo reported that the group is preparing a follow-up paper on implementing them (Building Trust in Agentic Commerce: Customer-centric principles (six banks, 2026); Source: newsroom.bankofamerica.com; gizmodo.com). On the merchant side, the paper's choice principle opposes "preferencing payment methods and other services (e.g. loyalty providers) at checkout" and "gatekeeping fees without reasonable commercial grounds", and asks that the identity and role of the merchant of record be clear and that consumers have the option to share their identity with merchants (Building Trust in Agentic Commerce: Customer-centric principles (six banks, 2026)). See Agentic Commerce.

Algorithmic pricing

Algorithmic and personalized pricing has become the most legislated area of retail AI. New York Algorithmic Pricing Disclosure Act (NY S 3008) is a mandatory disclosure law effective November 10, 2025; the National Retail Federation has mounted a First Amendment challenge to it. FPF — A Price to Pay: U.S. Lawmaker Efforts to Regulate Algorithmic Pricing documents 51 state bills in 24 states targeting algorithmic pricing. In August 2025, Delta Air Lines walked back AI ticket personalization under pressure.

Drone delivery

Walmart expanded drone delivery in June 2025 (WIRED), and Amazon has conducted ongoing drone testing. Federal Aviation Administration beyond-visual-line-of-sight (BVLOS) rules remain the regulatory constraint on broader deployment.

Deployment rollbacks

Several major retailers have retracted AI deployments after live operation revealed reliability shortfalls, providing concrete evidence for the enterprise AI deployment gap.

Starbucks retired its AI-powered "Automated Counting" inventory tool across North American stores effective May 18, 2026, roughly nine months after deploying it. The app scanned shelves using LIDAR and camera data to track stock but repeatedly miscounted and mislabeled items, for example confusing similar milk types, and Starbucks returned to standardized manual counting. The failure occurred in real-world accuracy on a visually ambiguous, high-volume inventory task rather than in the underlying perception technology in the abstract. (Source: https://www.reuters.com/business/starbucks-scraps-ai-inventory-tool-across-north-america-2026-05-21/)

Klarna reversed its AI customer service back to human hiring in May 2025 (Bloomberg).

Security and AI-enabled attacks

Online retailers were the target of an ongoing criminal campaign run largely by AI agents, documented in Gambit Security's interim report of September 22, 2026. The operator selected targets partly through a website traffic-ranking service, filtering out shops on the major hosted or open-source commerce platforms to keep those running custom code, and launched 105 attack projects between September 10 and 15 alone, compromising at least 27 companies to varying degrees. Gambit attributes to the campaign at least 600,000 unexpired credit card records from two companies, card-skimming scripts ordered against at least 27 named victims and confirmed on 19, and more than 100 further websites carrying an associated skimmer; the operator's own review put the mean cost at $25.46 per completed scan. At a bicycle retailer the agent's cleanup dropped 180 database tables, including backups the victim's administrators had made (Source: gambit.security). The campaign is described in more detail at Autonomous cyber-agents.

Relationships

• deployed-by: Walmart, Amazon, Target, eBay, Etsy, Wayfair, Klarna.
• regulated-by: FTC (consumer protection), state AGs (algorithmic pricing), FAA (drones).
• depends-on: OpenAI, Amazon, New York Algorithmic Pricing Disclosure Act (NY S 3008).
• related: Algorithmic Pricing and Antitrust, FPF — A Price to Pay: U.S. Lawmaker Efforts to Regulate Algorithmic Pricing, New York Algorithmic Pricing Regime and AG Enforcement, Enterprise AI Deployment Gap, Financial Services — AI Deployment (Klarna crossover).
• related: Agentic Commerce, Agentic Commerce Protocol (ACP), Building Trust in Agentic Commerce: Customer-centric principles (six banks, 2026) — agent purchasing and the banks' principles for it.