Report: AI Agents Are Becoming Retail’s New Gatekeepers

A new global report from AWS, Botify, DataDome and Retail Economics shows how agentic AI is rewriting search, discovery, and digital commerce. Retail search is no longer just about keywords.

According to a new research report, The Future of Search and Discovery: A Strategic Playbook to Understand Agentic Commerce, artificial intelligence systems are rapidly becoming the primary intermediaries between brands and consumers. Instead of shoppers typing product queries into search bars and scrolling through links, AI assistants increasingly interpret intent, retrieve data, compare options, and curate recommendations before a retailer’s website is ever visited.

The shift marks the emergence of what researchers call “agentic discovery” — a commerce environment where autonomous AI agents influence visibility, comparison, and even purchasing decisions.

And the transformation is already underway.

AI Adoption Is Already Mainstream

The report, based on a survey of 6,000 consumers across the United States, United Kingdom, and France, reveals that 73 percent of consumers have used AI assistants in the past year. Thirty-eight percent have used AI tools specifically for shopping-related tasks such as product comparisons, recommendations, or idea generation.

AI is no longer experimental. It is embedded into everyday search experiences.

Younger consumers are driving adoption. Roughly one in four shoppers aged 18 to 24 use AI assistants regularly, and one in five rely on them daily. Among consumers over 55, daily usage drops to fewer than one in ten.

The generational gap suggests that agentic commerce will accelerate as digital-native consumers gain spending power.

The Rise of Machine-First Discovery

One of the most dramatic findings in the report concerns AI-driven bot traffic.

AI bot activity increased 5.4 times over the course of 2025, fundamentally reshaping how retail websites are accessed and measured . Unlike traditional search engines that crawl primarily to index content, AI systems crawl intensively to ingest product data, validate pricing, and support real-time retrieval inside conversational interfaces.

This changes the economics of visibility.

Google generates one website visit for roughly every six crawls. OpenAI, by contrast, generates one visit per 198 crawls. In other words, much of AI-driven discovery and evaluation now happens upstream, inside AI interfaces, before traffic ever reaches a retailer’s domain.

Retail analytics dashboards that rely on clicks and impressions may no longer tell the full story.

The “Invisible Brand” Problem

AI systems cannot recommend what they cannot see.

The report highlights a structural weakness across modern retail websites: heavy reliance on JavaScript-rendered content. Many AI crawlers cannot fully interpret JavaScript, meaning product descriptions, specifications, and pricing details may be partially invisible to machine agents .

If product data is not structured, machine-readable, and accessible at the server level, AI systems may instead source information from third-party marketplaces, review sites, or even competitors.

In agentic commerce, technical architecture becomes brand strategy.

Security Risks Are Growing Alongside Opportunity

The rise of AI traffic brings visibility benefits but also new vulnerabilities.

DataDome analyzed 698,214 live websites using a spoofed AI assistant user-agent and found that nearly 80 percent did not block or challenge the impersonation attempt.

That leaves retailers exposed to distorted analytics, inflated referral signals, pricing exploitation, and automated fraud.

The challenge is distinguishing legitimate AI agents from malicious bots. Many AI systems do not clearly declare themselves, making verification difficult and blurring the line between beneficial crawling and exploitative scraping.

Retailers must now treat AI governance as both a marketing and security priority.

Trust Remains a Friction Point

Consumers are experimenting with AI in shopping journeys, but delegation remains limited.

While 38 percent say they feel comfortable receiving AI-generated recommendations, far fewer are willing to let AI act autonomously on their behalf . Thirty-two percent of consumers across surveyed regions say they do not trust AI-enabled search and discovery.

Nearly half of shoppers still prefer to conduct product discovery themselves rather than outsource it entirely to automation.

Agentic commerce is advancing faster than consumer trust.

Categories Most Exposed to Disruption

Not all retail sectors will feel the impact equally.

Electronics and appliances rank highest in consumer willingness to use AI for comparison and decision support . High-spend, high-complexity categories are especially vulnerable to AI mediation because shoppers value clarity, specification breakdowns, and price transparency.

Meanwhile, sectors with volatile pricing and inventory dynamics are experiencing disproportionate AI crawl intensity. Food and grocery websites, for example, saw AI-driven bot traffic increase 29 times over the year.

AI systems treat categories differently depending on data volatility and comparison complexity.

The Death of the Click

Perhaps the most profound implication is measurement.

Traditional SEO metrics were built for a click-driven internet. Agentic discovery reduces reliance on direct website visits. AI assistants summarize, compare, and recommend products within their own interfaces.

The report argues that retailers must develop new metrics such as:

• Agent inclusion rates

• Discovery visibility scores

• Trust signals

• Visibility-to-sale ratios

  
  

Without updated performance frameworks, brands risk optimizing for traffic that no longer reflects real influence.

A New Consumer-Retailer Dynamic

The research introduces what it calls an “Agentic Discovery Systems Framework,” built around three pillars: visibility, control, and trust.

Retailers must:

• Identify and verify AI traffic

• Structure and govern product data for machine readability

• Define access policies for automated systems

• Continuously assess trust at the transaction level

  
  

Discovery is no longer a linear funnel. It is a mediated system of AI answer engines, conversational interfaces, agentic browsers, and autonomous recommendation systems sitting between brands and customers.

Commerce in the Age of AI Intermediaries

Search once meant typing keywords into Google. Then it meant scrolling social feeds. Now it increasingly means asking an AI assistant for the “best waterproof hiking shoes under $150” and accepting a shortlist generated in seconds.

Retail visibility is shifting from ranking on a results page to being included in an AI-generated answer.

The brands that thrive in this environment will not just optimize for humans. They will optimize for machines.

And in agentic commerce, machines are becoming the first customer.