Read the State of Commerce 2026
Agentic AI and agentic commerce are hot topics right now, and with good reason. After all, the technology is fundamentally changing how today’s consumers shop online, shifting who is doing most of the work leading up to the order and redefining what a “normal” purchase journey looks like to merchants.
Considering agentic AI is projected to influence hundreds of billions in ecommerce sales by 2030 in the U.S. alone, it’s important to understand why agentic AI has struck such a chord with online shoppers today and what it could mean for your business tomorro
TL;DR
- Agentic AI is reshaping online shopping by solving decision fatigue, saving time and helping shoppers find better prices.
- Consumers are already using AI agents for low-risk shopping tasks like product research, deal hunting and cart building, with trust gradually expanding to higher-stakes purchases like insurance and travel.
- Some signals of agent-led shopping activity merchants should look out for include: short sessions with higher conversion rates, unnaturally fast/efficient browsing that hits lots of product description pages (PDPs), thinner device IDs and more AI-chat mentions.
Why shoppers are turning to agentic AI
The reasons more shoppers are leveraging agentic AI comes down to three things: decision fatigue, time and money.
Online shopping has always involved a lot of leg work on the customer’s side. They hop between ecommerce store websites, compare products and prices, read reviews, add contenders to their cart and search for promos or coupons to make sure they’re getting the best price before they pay.
In this traditional ecommerce model, shoppers must be researchers and analysts, before buyers. Every new tab they open to read one more review or compare another product is a withdrawal from their limited bank of mental energy. And every hour spent comparison shopping or hunting for coupon codes is time they’ll never get back. This cognitive load often contributes to abandoned carts — not because the customer didn’t want the item, but because finding the ‘best’ deal became more exhausting than the item was worth.
This taps into a fundamental principle of human psychology: People have a finite amount of mental energy for decisions each day. Why spend it comparing stand mixers when an agent can run that analysis in seconds and present you with the two best options based on your exact needs?
This is exactly why agentic AI is so captivating to consumers. With agentic commerce, autonomous agents take the time-consuming burden of research and discovery off the customer’s shoulders. By delegating those tasks to an agentic shopper, the shopper offloads the stress of finding the “perfect” option and strips out the most tiring parts of the process. What used to be a mountain of open tabs becomes a short set of “yes” or “no” decisions, with the best price in reach, minus the mental exhaustion.
What tasks do customers trust AI agents to handle today?
While Adobe found that 38% of U.S. consumers are already leveraging AI for help with their online shopping tasks, and those using AI for online shopping are more confident in their purchases, most aren’t handing over full control. At least, not yet. Why? Because trust is earned gradually, starting with low-risk tasks.
Researching and comparing items
Instead of spending time bouncing between search results, category pages and “best of” lists, shoppers ask agents to surface a short set of options that fit within their guardrails — which are usually around price range, use case, size, delivery windows and reviews. From there, the agent performs the initial research, compares products, summarizes buyer reviews and, based on the findings, helps the shopper narrow down the best options.
A shopper looking for a high-traffic area rug, for instance, asks an agent to compare two finalists. Instead of just listing prices, the agent summarizes the trade-offs: “Both rugs fit your dimensions, but the wool option is rated 20% higher for durability. However, the synthetic rug is machine washable and $150 cheaper, though reviews mention it ‘sheds significantly’ in the first month.”
Finding the best deals
According to a study by Wildfire Systems, 86% of ecommerce shoppers say saving money is a motivator to use agentic AI agents, with 46% specifically interested in getting help finding the best prices and deals, including gathering valid coupon codes or active promotions. In fact, Signifyd’s State of Global Commerce 2026 report found that 50% of all Cyber Week orders in 2025 included a discount code, signaling not only heightened consumer demand for value but also increasing agent activity in uncovering deals.
Instead of deal hunting manually, shoppers are also instructing agents to “watch” items for them and send an alert when:
- A price drops below a set threshold
- A new promotion goes live
For example, someone planning a three‑day trip could ask an agent to monitor nonstop flights from Chicago to Miami, alert them when a fare with a carry‑on included dips under their set price and highlight any fare rules before they book.
Making shopping lists and building carts
Customers are also saving time by asking agents to plan their purchases (“What ingredients do I need to buy to make a chicken pot pie from scratch?”) and build carts for them based on a set of guidelines, like budget limits, size and fit requirements, preferred brands, dietary restrictions or past order history.
Today, some customers are also letting agents handle low-risk checkouts. In fact, 20% of customers surveyed by Wildfire Systems said they feel very comfortable letting AI agents find grocery or household basics, add them to cart, confirm the order price and initiate check out for them.
Visualizing products before buying
While some retailers have offered AI-powered visualization tools for years — like Wayfair’s Decorify platform (formerly Muse) or Zenni Optical’s Virtual Try-On — today’s shoppers are using agentic AI to do this work for them inside the agent experience itself rather than navigating to individual merchant sites. They upload a photo of themselves (or their space) and ask the agent to generate an image showing how something will actually look in real life, like how a jacket might fit or if a wallpaper matches their room’s aesthetic.
What will consumers trust AI agents to do moving forward?
As confidence in the technology grows, customers will likely delegate even more of the end-to-end journey to AI agents, further changing their shopping behavior.
“Right now, most shoppers are using AI agents as research assistants - comparing products, summarizing reviews, hunting for deals. That’s the safe zone. The next shift is operational. Agents won’t just help decide what to buy, they’ll start managing the entire purchase lifecycle.”
Xavi Sheikrojan, Signifyd director, risk intelligence.
Proactive replenishment
The most immediate leap in trust will be for the things everyday customers are burdened to think about: the essentials. Instead of a shopper needing to remember to buy more cat litter or trash bags, an agent can monitor usage patterns and purchase history to reorder the items automatically. This moves the customer from a cycle of “Dang, I’m out” moments to a state of constant supply.
Higher stakes agent delegation
Once customers get completely comfortable with agents handling shopping tasks (and see them get it right consistently), the trust barrier for higher-risk, high-consideration tasks will start to dissipate. Agents will soon tackle the “bigger fish” categories like mortgage refinances, insurance policies, complex travel packages and even healthcare.
In these sectors, the best deal rarely shows up in the headline price; the real value (or pitfalls) is buried in the terms, exclusions and fine print, which can be time-consuming to uncover and murky even when shoppers find them.
Agents solve this by running thousands of “what-if” simulations against a contract’s terms to calculate the true cost of ownership. By auditing 100-page disclosure documents in milliseconds or cross-referencing a new medical plan against a user’s actual health records to flag coverage gaps, these agents will act as a verified safety net.
“Once trust builds in these platforms, and once the agents are fully integrated into wallets and lifestyles, we can expect shoppers to rely on agents to proactively hunt for and retrieve 'bigger fish,' like mortgage refinances, car insurance policies and even medical procedures. We’re already seeing platform bets that assume users will do exactly that, with Open AI recently announcing ChatGPT Health as an example.”
Vito Petruzzelli, Signifyd senior risk intelligence analyst
Post-purchase mediation and returns
For many, the most draining part of online shopping is the return process: digging through emails for order numbers, confirming they meet the ecommerce retailer’s return policy requirements, navigating clunky retailer portals and waiting on hold for approval, hoping for a fast, or instant, refund.
In the future, shoppers will trust agents to negotiate the back-end of the transaction, acting as a personal logistics manager: initiating returns with a simple “this jacket doesn’t fit,” uploading photos or measurements automatically, reviewing the retailer’s return policy and even disputing partial charges if the item arrived damaged.
Embedded financial optimization
Over time, agents will also optimize around financial outcomes, not just convenience. As trust deepens, the agent won’t just decide what to buy — it will decide how to pay. Agents will choose payment methods based on approval likelihood, rewards optimization or installment structures. They may delay purchases if credit utilization is high or if better promotional offers are likely to appear soon.
“In the future, the agent will become part financial planner, part procurement engine — dynamically balancing liquidity, credit health and long-term value while executing transactions on the user’s behalf.”
Xavi Sheikrojan, Signifyd director, risk intelligence
Full trust won’t happen overnight, but friction will eventually beat fear. Just as consumers moved from “I’ll never put my credit card on the internet” to “Amazon One-Click,” they’ll eventually find the manual checkout process so tedious that they’ll delegate it to an agent by default. That is, once the agent has earned enough trust.
“Twenty years ago, people wouldn’t want to put their credit card on the internet because they were worried and now, look. We don’t even blink.”
Carl Boutet, founder at StudioRx
Once agents have shown their trustworthiness, shoppers will delegate the end-to-end shopping journey to them — they’ll handle finding products, adding them to cart, searching for and applying coupons or promos, adding shipping details, using a saved payment method and placing the final order. The agent will only ask the human customer for a decision when it comes to things like setting up a subscription or updating the payment method on file if it failed.
How agentic AI is already affecting your business
Realistically, agentic AI is already reshaping your customer’s journey, even if it hasn’t shown up as a glaring channel in your dashboards yet. Here are a few signs to watch for that indicate agents are already influencing how shoppers discover, evaluate and buy from you.
The top agent fingerprints to look for in your analytics
- Ultra-qualified visitors: When you look through your data, you may notice a segment of traffic with a significantly higher conversion rate than your regular site average. This often represents shoppers who have already been “vetted” by an agent. When they arrive, they skip the home page and go straight to a specific product description page or PDP.
- Silent audits: Look for session behavior that is “unnaturally” efficient. While humans tend to wander around websites as they shop, agents move with clear intent — hitting multiple PDPs, reviewing return policies and confirming stock levels in seconds. If you see high-volume traffic that heads straight for your structured data (like your JSON-LD or schema files) while ignoring visuals that would make human browsers pause, that’s agent activity.
- Less mobile-based traffic: Device information is notoriously absent when agentic agents are involved, so take a look at your device mix. If you see fewer mobile device IDs and more sessions that look generic, virtualized or stripped of identifiers, those sessions were agent-led.
- More IP addresses tied to where big data centers are: A clear signal of agentic activity is when a higher share of sessions starts originating from cloud regions and data-center hubs (like Northern Virginia) rather than residential areas or mobile networks. In other words, the “where” in your logs increasingly reflects where the agent runs, not where the shopper lives.
- “Zero-click” discovery: Check your brand mentions in AI-powered chats, like Gemini, ChatGPT or Copilot, and look at your website referral traffic. If your products are being recommended in conversational interfaces and you’re getting referral traffic from AI chatbots and tools but your click-through rate (CTR) is flat, this signals that agents are using your product information to make recommendations within their interface.
What agentic AI behavior really means for merchants
Agentic AI doesn’t just change where discovery happens. It changes what a “normal” shopping session looks like, not to mention what you can rely on to guide decisions across marketing, merchandising, fraud and customer experience.
First, expect fewer browsing signals and more “ready-to-act” traffic. When an agent does the comparison work upstream, shoppers arrive later in the funnel. That can show up as higher conversion rates on certain SKUs, shorter sessions, fewer pageviews and less engagement with the site elements (like banners and category pages) you’ve usually used to steer human buyers.
Second, prepare for messier attribution and a new kind of visibility problem. If your products are getting recommended in agent-led interfaces, you may feel the impact without seeing a proportional lift in CTRs. Mentions, citations and zero-click influence start to matter more because the agent layer can filter options before a shopper lands on your website.
Third, recognize that agents will change the shape of risk. This isn’t because agents are inherently bad, but because they abstract signals fraud systems have historically leaned on. If more sessions are happening without familiar device identifiers or move at “too efficient” speeds, static fraud rules built around known patterns will likely misfire.
For example, an agent compares five high-powered blenders and recommends the best option. The shopper clicks through to that specific product page and checks out in under 20 seconds.
To a legacy model, that can look identical to an automated bot run — minimal browsing, thin device data and fast path to purchase — and the order gets blocked. To a model trained on agent-led behavior, however, that same pattern reads as genuine instead of agentic fraud, and moves it along.
Turn agent-led activity into trusted commerce transactions
As agents change how people shop, merchants can’t rely on the same fraud signals from a year or six months ago. You need systems that recognize agent-led shoppers for what they are: high-intent customers, not threats.
Signifyd’s Commerce Protection Platform does exactly that. Using machine learning trained on evolving shopping behavior and fraud patterns, our platform separates legitimate agent-led activity from bad bots, identifying intent even when device data and browsing behavior no longer tell the full story. Request a demo to see how.
Photo by Getty Images
Want to dive deeper into agentic commerce and how it’s shaping 2026 ecommerce trends? Read our Global State of Commerce 2026 report.
