Say you lose your headphones. You were happy with them, but you’re also open to switching brands. That once called for a visit to various retailers’ websites to start the arduous product comparison process, like the pilgrims did. With agentic commerce, you can delegate that task to an AI agent. The agent can find the best deal, apply promo codes, and even check out on your behalf.
Naturally, agentic commerce was the hottest topic at the National Retail Federation’s recent Big Show, where Google announced Universal Commerce Protocol (UCP), which Forter now supports. UCP will power a new checkout feature on eligible Google product listings in AI Mode and the Gemini app. Shoppers can securely make purchases as they’re researching.
That’s great news for merchants that work with Google, who retain control over the customer experience. It’s also great news for merchants that work with Forter. Read on to learn why.
The Agentic Future Is Now the Agentic Present
Unlike rule-based bots or simple scripts, agentic AI can act independently. With AI-assisted shopping, agents browse and compare products, but stop short of purchasing. Humans must complete the transaction. Fully autonomous AI agents can browse and buy, functioning as a proxy for a human user — examples include Perplexity Pro and OpenAI’s Instant Checkout.
That’s great for customers, who can interact with brands the way they want to. It’s been more of a mixed bag for merchants, who have lacked visibility and control of the customer experience.
At a media lunch we hosted in October, Brooklinen COO Rachel Levy pointed out that AI referral traffic lacks context. “[Did the customer search] ‘the Internet’s favorite sheets? We don’t know, and we probably need to do some consumer research to say, ‘Why did you start your journey here?’ to understand,” she said.
When shoppers make purchases with UCP, the retailers remain the merchants of record. They preserve the customer relationship with full visibility into transaction data, loyalty programs, and post-purchase support, such as returns and exchanges. That puts control back in their hands.
Agentic Commerce Ushers in a New Kind of Fraud
Agentic referrals are visits to a merchant’s site from platforms like ChatGPT, Gemini, Claude, or Perplexity. And they’ve exploded in recent months. During Cyber Month, we saw a 40% increase in AI referrals. These visits were most prevalent in beauty, fast fashion, tickets, and travel. We also saw a 23% increase in AI referral conversions, nearly half of which were driven by first-time users.
However, this spike in agentic traffic doesn’t just mean more sales — it also ushers in a new, different kind of fraud. Fraudulent agentic transactions increased by 970% in the last six months alone.
Some merchants’ kneejerk reaction may be to place a blanket ban on bot traffic. Others have long had such a ban in place. It wasn’t long ago that bot traffic was universally considered bad news. That legacy approach won’t work in the agentic future.
Most of the bot traffic we saw during Cyber Month was malicious, designed to stack promo codes, steal data, or commit return fraud. But 10% of that traffic came from legitimate customers. Had merchants blocked all bot traffic, those sales wouldn’t have happened. False declines also have a downstream effect of alienating good, and possibly loyal, customers.
Without identity intelligence, merchants can’t distinguish the two. If they can’t tell the difference between a good and a bad bot, they’ll forever be playing catch up with a technology they can never out run.
Getting Your Brand Ready for Agentic Commerce
Getting your brand agentic commerce-ready starts with some questions. The answers will help assess your readiness, while uncovering any gaps.
- Are our risk systems built for agentic behavior? Legacy fraud models are based on human behavior. AI agents complete tasks far more quickly, and they scroll and enter data differently, which makes it easy for them to be misclassified as fraud.
- Can we distinguish helpful automation from abuse? Without advanced identity intelligence, AI agents and malicious bots may look identical. Identity also creates a holistic customer view, which allows you to recognize an AI agent purchasing from three different sites as the same account.
- Are teams aligned on how automation shows up across the journey? Fraud, payments, and digital are often dispersed among three separate teams. Agentic AI doesn’t fit cleanly in just one of those buckets. For example, the fraud team may remove a CAPTCHA to improve approvals for AI-led orders. The digital team can reintroduce it for “security.” Those blurry lines make frictionless flows, which agents need to function properly, impossible.
- Is personalization logic tied to verified identity or vulnerable to spoofing? This is another area where identity is crucial. Spoofed agent profiles may trigger VIP discounts they’re not actually entitled to receive.
Steps Merchants Can Take to Prepare for Agentic Commerce
Once you’re clear on the answers, there are steps to take across data, operations, technology, and the customer experience.
Data readiness:
- Product data enrichment: Ensure all product data (descriptions, specifications, images) is complete, accurate, and structured. This includes detailed metadata for features and attributes. AI doesn’t reason the way humans do, so wrong data results in confidently wrong actions.
- Structured data and schema markup: Implement and validate up-to-date schema.org markup (Product, Offer, Review) on all product pages. This makes data machine-readable.
- Inventory and pricing feeds: Maintain real-time, high-fidelity feeds for inventory levels and dynamic pricing. Agents require immediate and trustworthy information.
Operational readiness:
- Fulfillment and logistics: Establish clearly defined Service Level Agreements for shipping and returns, making them readily accessible to agents. Automated processes for shipping and returns should be implemented as well. AI agents prioritize merchants with minimal friction.
- Agent-optimized communication: Develop communication pathways specifically designed for agent interaction rather than human customer service. This includes dedicated APIs for status checks, order modifications, and basic inquiries such as “Where is my order?” You should also have a machine-readable knowledge base (using JSON-LD or similar) containing answers to common agent queries.
Tech readiness:
- API ecosystem development: Build robust, well-documented, and secure APIs for all transactional touchpoints: search, checkout, order management.
- Scalability Testing: Stress-test systems to handle potential surges in transaction volume driven by agents.
- Security and Trust: Implement strong authentication protocols, such as OAuth 2.0, and data security measures, as agents are programmed to assess security profiles.
- Platform integration audit: Review compatibility with major agent platforms, ensuring seamless data exchange and checkout flows.
Customer experience readiness:
- Review and rating visibility: Ensure product and merchant reviews are easy to parse and highly visible in structured data. Agents will compare star ratings and review volume.
- Policy clarity: Clearly articulate all consumer-facing policies, such as shipping costs, guarantees, environmental impact data.
- Agent-specific offers: Ensure promotions and offers are targeted toward the logic of purchasing agents. This includes machine-discoverable dynamic bundles based on purchase history or pre-approved decision trees that allow AI agents to engage in pre-defined minor negotiations on price or shipping speed.
How Forter Helps Merchants Be “Agentic Commerce Ready”
AI has always been the center of Forter’s fraud prevention. As it’s become enmeshed in the customer journey, we’ve introduced new capabilities to our platform, designed to help merchants thrive in the agentic era.
Our solutions are powered by the world’s most powerful identity network, comprising more than 2 billion unique online identities. Each identity includes 6,000 behavioral and hard data attributes. This allows us to precisely identify browsing and shopping behaviors, and link them to the person behind the bot. And it allows you to block malicious agentic traffic without adding friction for legitimate customers.
Our Trusted Agentic Commerce Protocol is also designed to solve the data and trust challenges that make merchants wary of agentic commerce. An open standard, the protocol allows both merchants and agent developers to authenticate each other, maintain rich customer data, and prevent fraud, while putting the customer first.
Interested in learning more? Request a demo.
