The ecommerce landscape is on the brink of its most dramatic transformation since the shift from desktop to mobile. As we’ve witnessed the explosive advancement of generative AI over the past year — from ChatGPT o3’s deep research capabilities to Claude Opus 4’s superior coding performance — we’re rapidly approaching a future where AI agents don’t just assist with shopping, but actively shop on our behalf.

For merchants, this isn’t a distant possibility; it’s an imminent reality that will fundamentally reshape how business is conducted online.

The current AI shopping revolution

We’re already seeing the early stages of this transformation. Consumers are increasingly using large language models like ChatGPT, Gemini, and Perplexity to research products, compare prices, and discover deals. While true autonomous shopping agents haven’t reached mainstream adoption yet, the infrastructure is rapidly falling into place. Major payment networks like Visa and Mastercard are working directly with OpenAI and other AI companies to enable secure agent-driven transactions, complete with spending limits and tokenized credentials.

ChatGPT can already show you product recommendations with direct purchase links — we’re just steps away from having an agent complete those purchases on your behalf.

Three possible futures

The trajectory of agentic commerce is likely to unfold in one of three scepnarios, each with dramatically different implications for merchants:

  • Scenario 1: Merchant-controlled ecosystem In this future, merchants successfully maintain control by keeping AI shopping native to their platforms while implementing sophisticated barriers against external agent access. Picture Amazon’s AI shopping assistant or Target’s native agent. Merchants retain maximum control over customer relationships and pricing power, but must invest heavily in their own AI capabilities to remain competitive.
  • Scenario 2: Collaborative ecosystem In this scenario, a framework emerges where merchants and shopping agents work together through standardized protocols. Agents access merchant inventory through secure APIs, while merchants maintain some pricing flexibility and brand presence. This creates a balance where merchants retain partial control while capturing agent-driven traffic.
  • Scenario 3: Decentralized shopping Merchants should be most worried about this scenario. In this world, shopping agents become the dominant interface for commerce, reducing merchants to mere inventory suppliers competing primarily on price and fulfillment speed. Brand differentiation becomes nearly impossible as merchants are forced to participate or risk losing enormous traffic volumes.

The strategic dilemma

Each future presents merchants with the same fundamental tension: cooperate with agentic traffic and risk losing control of the full customer journey, or resist and potentially lose relevance entirely.

In the disintermediated scenario, merchants effectively become inventory catalogs within agent ecosystems. Without direct customer relationships, they lose the ability to differentiate, create brand loyalty, or control their destiny. This is similar to how airlines became commodity providers within online travel agent (OTA) platforms like Expedia — but potentially more challenging, as merchants would compete against hundreds of others selling identical goods.

The hidden fraud challenge

Beyond the strategic considerations lies an equally critical challenge that many merchants haven’t considered: fraud and abuse prevention in an agentic world.

Traditional fraud detection relies on behavioral signals like device fingerprinting, session tracking, browsing patterns, and user interactions. When AI agents make purchases, these signals either disappear or become misleading. Depending on how this ecosystem develops, merchants might have to assess order risk with only a few data points, like shipping address and phone number.

Early data from Riskified’s merchant network shows LLM-referred traffic to be more risky in certain industries. For example, LLM-referred traffic from a large ticketing merchant was 2.3x more risky compared to Google search traffic. In another example, an electronics merchant showed 1.8x riskier traffic. This trend will likely accelerate as bad actors deploy their own sophisticated agents to automate fraud at an unprecedented scale while maintaining human-like purchasing patterns.

Consider these emerging fraud vectors:

  • Agent takeovers: Compromising a customer’s ecommerce account and using the “Buy for Me” feature across multiple merchants
  • Automated reseller arbitrage: Deploying agents to systematically strip inventory faster than human scalpers, then optimizing marked-up storefronts to ensure other agents recommend their products
  • Sophisticated fraud scaling: Automating the entire fraud lifecycle from credential phishing to multi-merchant testing at massive scale

How to best prepare

Agentic shopping presents a pivotal moment for ecommerce, requiring merchants to balance innovation with security. To thrive in this new era, collaboration between risk professionals and business leaders is vital. 

With 13 years of expertise in fraud prevention, Riskified has helped merchants adapt to every shift in ecommerce and shopper behavior, from the transition to mobile to the impact of COVID, to the evolution of AI. By combining advanced machine learning with real-world insights from our merchant advisory groups, we empower businesses to outpace fraud trends and protect their revenue.

Riskified is uniquely positioned to guide businesses toward success in the near-future of agentic commerce, offering strategic advice, adaptive fraud models, and chargeback liability protection to safeguard growth through uncharted terrain. Traditional fraud prevention systems can’t meet these demands, but our solutions ensure merchants stay resilient and ahead of threats.

Partner with Riskified and secure your business in a rapidly changing ecommerce landscape.