AI agents are already mediating the relationship between brands and shoppers—and most businesses aren’t prepared for what agentic commerce means for their visibility, their data, or their security. 

That’s the message at the core of our recent webinar, The Future of Search & Discovery: Understanding Agentic Commerce, featuring insights from Jerome Segura, VP of Threat Research at DataDome, Richard Lim, CEO of Retail Economics, AJ Ghergich, VP of AI & Consulting Services at Botify, and Marco Kormann Rodrigues, Partner Leader of Retail and Consumer Goods at AWS. 

Together, they explored how agentic commerce is changing discovery, how businesses can ensure their products are visible to AI, ways to distinguish good AI from malicious automation, and more. Watch the full webinar replay here. 

Here are the top five takeaways from the session. 

1. Consumer AI adoption is already mainstream

The numbers don’t lie: AI-powered discovery is no longer an emerging trend; it’s the new normal. 

73% of consumers across the US, UK, and France have already used AI assistants for product discovery, with nearly 40% of consumers already using AI for shopping tasks. 

Interestingly, the data points to wealthy millennials as the group most likely to trust and use AI for shopping. With 97% of this group planning to continue using AI assistants for shopping, it isn’t just a fad either. 

2. The discovery funnel is fragmenting 

For years, the formula for retail visibility was relatively straightforward: strong website, solid SEO, and you’d be found. That playbook is breaking down.

Today, answer engines, shopping agents, and agentic browsers all mediate the relationship between consumers and a brand ahead of a purchase. Purchasing journeys are collapsing from 10 steps to one or two, and happening entirely inside an AI chatbot interface.

For e-commerce businesses, this marks a shift in focus from channel ownership to data control, data signals, and technical accessibility. Providing AI agents with structured, accessible data that provides enough context on your products is necessary for visibility: 

“Messy, inconsistent data and taxonomies kind of render your brand invisible. The data is huge,” AJ explained.

3. AI bot traffic distorts analytics & makes measurement difficult

As AI agents evaluate products on behalf of consumers before they ever visit a site, traditional engagement metrics such as clicks, sessions, and pageviews are becoming increasingly unreliable proxies for brand presence and performance. 

With AI bot traffic increasing 5.4x in 2025, the scale of this shift is hard to ignore. AI systems crawl more intensively than traditional search: where a Google crawl generates roughly one visit per six crawls, AI-driven discovery generates approximately one visit per 198 crawls—creating significant infrastructure load while diluting website analytics.

The problem is compounded by agents that don’t follow established standards. Jerome cited Grok from xAI as an example: a single page request triggered 12 simultaneous requests from different IP addresses and user agents, causing one interaction to register as many distinct sessions in your logs.

This is why, in the age of agentic commerce, bot management isn’t only about stopping clearly malicious attacks: 

“Part of the work we do at DataDome is not always to block these malicious attacks with malicious intent, but also to filter and block these requests that are simply abusive—that don’t really meet the business interest,” Jerome said. 

An agent that hammers your infrastructure with redundant requests, distorts your analytics, and provides no legitimate value to your business falls into that category, regardless of whether its intentions are explicitly harmful. 

Without the ability to distinguish between human, bot, and AI agent traffic and its intent, you can’t accurately assess your threat surface or make confident, data-driven business decisions.

4. Not all AI agents are created equal

Managing AI agent traffic isn’t just about volume. It’s about knowing who, or what, is actually at the door—and what they’re trying to do. Right now, that visibility is severely lacking. 80% of AI agents do not properly identify themselves, and 80% of websites do not verify agent identity.

The result is a system built largely on trust that is easily exploited by fraudsters:

“We did some simple spoofing testing, where you use something very common and very abused—your user agent string—saying, ‘Hey, I’m a ChatGPT user.’ And what we found is most websites actually just believe you at that. That means either they don’t have a bot protection solution in place, or they fear that blocking traffic from any AI agent could impact their business,” Jerome explained.

That fear is understandable, but acting on it by leaving the door open to all self-declared AI agents creates serious exposure. A bad actor simply needs to claim the right identity to gain access.

This is why a binary “block all / allow all” approach to AI agents no longer holds up. Security teams need granular, intent-based bot and agent trust management that can distinguish between a legitimate indexing crawler, a bad actor hiding behind a spoofed identity, and an agentic browser acting on a real consumer’s behalf—each of which warrants a different response.

5. Now is the time to act

The fully autonomous purchase journey, where an AI agent completes a transaction end-to-end on a consumer’s behalf, is still emerging. But discovery and research are already agentic for a large share of consumers today. 

For retailers, that means three workstreams need to be in motion now: ensuring your product data is structured, consistent, and machine-readable; building on-site AI experiences that meet rising consumer expectations; and establishing clear traffic policies that let legitimate AI agents in while keeping abusive ones out. 

For security teams, the urgency is just as high. The threat surface is expanding quickly and in ways that traditional tools weren’t designed to handle—from spoofed agent identities to agentic browsers that look indistinguishable from human sessions. The window to build the right foundations is open now.

As Richard Lim, CEO of Retail Economics, put it, “The businesses that are going to win in this new environment are those that are focusing on all of these areas—not in isolation, but as a coordinated strategy.”

Watch the replay to learn more

Agentic commerce isn’t a future state. It’s here now and adoption is trending up. The question is no longer whether AI agents will interact with your business, but whether you’re ready for them.

If you missed the live session, watch the replay here, or download the Future of Search & Discovery Report for an in-depth look at the research that informed the webinar. 

Looking to gain visibility and control over your agentic traffic? Book a demo to learn how DataDome can help.