Agentic Commerce — AI Buys On Your Behalf

Agentic commerce is the model in which autonomous AI agents act as proxies for buyers — executing the full purchase lifecycle (discovery, authorization, payment, fulfillment) from a goal rather than a click. A user says “order running shoes under £90 that arrive by Friday,” and the agent evaluates options across merchants and completes the transaction, within preset spending limits and preferences.

Concrete examples live: ChatGPT’s “Buy it in ChatGPT” feature is live for US users; Google launched “Buy for me” buttons across AI Mode and Gemini; Walmart integrated its Sparky assistant into ChatGPT and reports roughly 2× as many new customers versus traditional search channels. This is not a roadmap item — it’s operational.

The implication most brands miss: If your product data isn’t machine-readable and structured, agents won’t surface you. They don’t browse — they query APIs and parse structured data.

Conversational AI Shopping Assistants Replace Search Bars

The shift from keyword search to conversational discovery has crossed a critical adoption threshold. In many workflows, shoppers no longer type “blue running shoes size 10” — they tell an AI “I need shoes for a half-marathon in October, I overpronate, budget around £100.”

Amazon’s Rufus (powered by custom LLMs, processing tens of millions of queries with under 300ms latency), Walmart’s Sparky, and Sephora’s ChatGPT-integrated assistant are all live examples. These aren’t chatbots with a shopping skin — they’re goal-oriented assistants that understand context, compare products, and guide the full decision.

What gets wrong: Most brands are still optimizing for keyword search. The new optimization target is AI recommendation retrieval — whether your product shows up when an AI is asked a relevant intent question.

The Machine-Readable Commerce Layer

As agents do the shopping, they depend entirely on structured, consistent, real-time data. This is driving a rapid shift toward what infrastructure teams call the “machine-readable commerce layer”: exposing product, pricing, and inventory data via APIs; ensuring schema.org Product markup on every PDP; and maintaining completeness across all product attributes.

Agents do not guess. If your product lacks proper markup or has inconsistent attribute data, the agent moves to a competitor that doesn’t. Retailers who invested in this infrastructure in 2025 are now capturing disproportionate agent-driven traffic.

Agentic B2B Procurement

The agentic shift isn’t only consumer-facing. In B2B, AI agents are automating complex purchasing, inventory renewal, and invoice reconciliation — connecting with vendor systems, managing supply chain procedures, and in some workflows, negotiating prices autonomously. Forrester predicts that 1 in 5 sellers will be compelled to respond to AI-powered buyer agents with dynamically delivered counteroffers via their own seller-controlled agents. B2B commerce is heading toward a world of agent-to-agent negotiation.

AI-Powered Price Discovery Reshapes Deal-Finding

AI-powered assistants can now scan a broad universe of online retailers, promotions, and historical pricing in seconds. The consumer who used to spend 20 minutes on price comparison sites now gets that outcome in a conversational turn. In 2025, 17% of consumers using AI tools for online shopping reported finding meaningfully better deals as a result, with 20% more planning to adopt this behavior. Dynamic discounting and real-time price matching are no longer just competitive tactics — they’re table stakes for staying visible in AI-mediated discovery.

Real-Time, Per-Person Decision Engines

The distinction that matters: most “personalization” today is still segmentation. Many shoppers see the same hero banner because they share a demographic trait or browsed the same category last week. Hyper-personalization works differently — it makes experience decisions at the individual level, in real time, using behavioral signals, contextual data (device, location, time), and business constraints (inventory availability, margin targets).

49% of Americans say AI recommendations affect their purchases. 64% say they’d be willing to buy products suggested by generative AI. The infrastructure requirement is non-trivial: unified customer data, real-time identity resolution, and an omnichannel activation layer that can apply the same decision logic across email, on-site, and paid media simultaneously.

AI-Generated Dynamic Product Content

Beyond recommendations, generative AI is now creating product descriptions, email copy, and campaign creative in real time — dynamically adjusted per user segment, per channel, and per context. A product page for a camping tent might automatically render technical spec copy for an experienced hiker versus benefit-led lifestyle copy for a first-time buyer, using the same underlying product data.

McKinsey estimates generative AI in retail could generate $400–660 billion in annual value. In many workflows, AI-driven email marketing is improving click-through rates substantially versus batch-and-blast approaches.

Predictive Personalization — AI Anticipates Needs

The frontier of personalization isn’t responding to what customers want — it’s knowing before they do. Subscription reorder modeling, churn prediction, and contextual triggers (weather, local events, seasonal patterns) now combine to surface offers before a customer has consciously formed an intent. Retailers describe this as moving from reactive recommendation to proactive anticipation. The key input is behavioral data richness; the key risk is creating an experience that feels intrusive rather than helpful.

Fashion AI: 50% of Purchases Influenced by Personalization

Fashion is the vertical where AI personalization is deepest and the data most striking. Style preferences, seasonal trends, and visual merchandising requirements have made fashion an early proving ground for AI-driven commerce. Beauty sees near-universal success from personalization investment — 94% of beauty marketers report sales improvement. The pattern: visual, preference-driven, high-SKU categories benefit most from AI personalization.

Omnichannel Personalization Coherence

A common failure mode in personalization: the AI makes a great recommendation on-site, then the customer sees a completely different message in their email the same morning. In 2026, leading retailers are running personalization from a single decision engine that propagates across all channels — owned and paid. The concept: one customer profile, one decision system, multiple activation surfaces. Without this coherence, personalization efforts often cannibalize each other or create contradictory experiences that erode trust.

Visual Search Becomes a Primary Discovery Channel

Visual search has matured from a novelty into a genuine conversion driver. The workflow: a user photographs a product they see in the street, a social post, or a friend’s home — and AI matches it to purchasable inventory instantly. Retail leaders point to multimodal visual discovery as one of three accelerating trends reshaping how consumers discover products. The conversion advantage is significant; shoppers using visual search often arrive with higher intent than keyword browsers because they already know exactly what they want.

Virtual Try-On & AR Product Preview

The returns problem in fashion and beauty is expensive — roughly 20–30% of online apparel orders are returned. AI-powered virtual try-on directly attacks this cost. By 2026, 91% of Gen Z shoppers express interest in AR shopping experiences, and retailers report conversion rate improvements of up to 94% from 3D and AR product visualization. The technology is no longer experimental; it’s live at scale with major beauty and fashion brands. The remaining barrier is production effort — creating 3D assets for large catalogs is still costly.

Social Commerce + AI on Roblox and TikTok

Social platforms have become primary shopping destinations, not just awareness channels. Roblox is now the fastest-growing Gen Z commerce channel by order volume growth — outpacing TikTok — according to research from Retail Technology Show 2026. AI is the connective tissue: recommendation engines that understand social context, conversational commerce embedded in social feeds, and AI-generated product experiences within gaming environments. Any brand with a Gen Z audience needs a position on this.

AI-Powered Merchandising That Reacts to Context

AI-powered merchandising platforms now integrate hyper-local data — weather shifts, regional events, social media trends — to dynamically adjust product placements and promotions. A retailer in Manchester sees different front-page merchandising than one in Nice, not because a merchandiser manually configured it, but because the AI has read local signals and reweighted inventory visibility accordingly. This removes the “one-size-fits-all” storefront problem that has plagued national retailers for decades.

AEO — AI Engine Optimization Replaces Pure SEO

The question “does Google rank my product?” is being joined — and in some segments, replaced — by “does the AI recommend my product?” Brand visibility in an agent-mediated world depends on AI optimization: whether your product shows up when a shopping agent is evaluating options. This is driving a new discipline, sometimes called AEO (AI Engine Optimization), focused on product data completeness, structured markup, API accessibility, and prompt-optimized product descriptions. Where SEO was about keywords and links, AEO is about data structure and agent-readability.

AI Demand Forecasting — Zip-Code-Level Precision

Walmart’s AI-powered inventory system integrates historical sales data, weather patterns, macroeconomic indicators, and local demographics to predict demand at zip-code level. The system’s patent-pending “anomaly forgetting” capability excludes one-time events from forecasting models — preventing a single Black Friday from distorting future predictions. The result: 30% fewer stockouts across 4,700 stores while optimizing transportation routing. This is the kind of operational gain that moves margins by multiple percentage points.

Dynamic Pricing Moves from Airlines to All Retail

Dynamic pricing — long the domain of airlines and hotels — is becoming standard in general retail. 25% of major e-commerce platforms are already using AI-powered dynamic pricing to adjust prices in real time based on demand signals, competitor pricing, inventory levels, and customer segments. The opportunity: 5–10% margin improvements through intelligent pricing, according to data from Dynamic Yield. The risk: consumer perception. Shoppers who discover they paid more than a neighbor for the same product at the same moment will not forget it. Price personalization requires careful guardrails around fairness and transparency.

Autonomous Supply Chain Operations

NRF identified autonomous supply chains as one of 2025’s two breakthrough moments (alongside smart consumer agents). AI systems are now making multi-step logistics decisions autonomously: rerouting shipments around disruptions, adjusting safety stock levels in response to demand signals, and coordinating across suppliers without human intervention in the loop. The shift from “AI-assisted” to “AI-autonomous” in supply chain is well underway at major retailers — and the risk profile changes significantly when AI errors compound across a chain rather than a single decision point.

AI Fraud Detection in Agentic Transactions

As AI agents complete transactions autonomously, fraud detection faces a new challenge: distinguishing legitimate agent traffic from malicious agent traffic. 78% of financial institutions expect fraud to increase from AI shopping agents, and 2026 is seeing authentication frameworks begin to formalize. The implication for retailers: transaction logging must now capture agent identity, authorization context, and decision signals — not just payment credentials. Retailers investing in agent-aware security infrastructure are better positioned for both risk management and compliance.

AI Customer Service — Moving Beyond Chatbots

The first generation of AI customer service was chatbots with decision trees and keyword matching. The current generation uses large language models that understand context, handle complex returns and complaints, escalate appropriately, and maintain consistent brand voice across millions of simultaneous interactions. The operational case is clear; the brand risk is also clear — LLM-based customer service can fail in ways that scripted bots never could, including off-brand responses and confident misinformation about policy details.

Physical Stores Becoming AI-Native Experience Centers

The stores that are winning in 2026 aren’t competing with e-commerce — they’re doing something e-commerce cannot. Brick-and-mortar is being redesigned as an AI-augmented experience layer: interactive AR displays, AI-assisted selling that uses real-time customer data, and “phygital” touchpoints where digital and physical merge. Pop-up retail is growing rapidly (projected from $95B in 2025 to $144B by 2032 globally) as brands use short-term immersive formats to create experiences worth the trip.

Invisible Checkout — Frictionless Payment

The next frontier of conversion optimization is removing checkout friction entirely. Walk-out technology (pioneered by Amazon Go), biometric payment, and pre-authorized agent wallets are converging on a future where “checkout” is simply confirmation of a transaction that already happened. The consumer benefit is obvious. The regulatory complexity — around authorization, data consent, and return rights — is equally significant, especially in the EU where consumer protection law is stringent.

AI-Powered Sustainability Verification

Sustainability claims are increasingly a conversion driver, particularly among under-35 shoppers. But greenwashing has eroded trust. AI systems that can verify and surface supply chain sustainability credentials in real time — integrating third-party certification data, carbon footprint calculations, and material sourcing records — are becoming a competitive differentiator for brands with genuine sustainability commitments. The risk: AI can also make verification seem present when it is not, amplifying greenwashing at scale if systems are not carefully audited.

Secondhand and Circular Commerce Powered by AI

Secondhand shopping continues to grow, and AI is making resale infrastructure viable at scale — automated grading of product condition, dynamic pricing for pre-owned items, matching buyers to sellers based on style preferences, and AI-generated product descriptions for one-of-a-kind items. NRF explicitly flags whether secondhand shopping will continue to thrive as one of 2026’s key unanswered questions, as the segment faces both growing demand and growing competition from AI-enabled resale platforms.

The EU AI Act Changes Everything for European Retail

The EU AI Act implementation in 2025–2026 is introducing compliance requirements that will significantly shape how AI shopping systems operate across Europe. High-risk AI applications — including some forms of personalization and automated decision-making — require transparency, human oversight mechanisms, and documentation. For retailers operating across borders, this creates a compliance bifurcation: systems designed for the US market may need meaningful rearchitecting for EU markets. This is not a distant concern — it’s an active compliance workstream for any retailer with European customers.