Michael Quoc holds 9 US patents and is the founder and CEO of Product.ai—a company building the truth layer for commerce.
Agentic commerce will be stalled until consumers can trust AI product recommendations.
Today’s commerce technology road maps ask whether an AI can complete a purchase. But they fail to ask if consumers will trust AI with their credit cards and their most critical shopping decisions.
Though McKinsey estimates agentic commerce as a multitrillion-dollar opportunity, Gartner found no more than 11% of shoppers were willing to let AI make a purchase decision on its own, even in low-stakes categories like household supplies.
The only way we’ll achieve the McKinsey numbers is to build agentic commerce tools that consumers can believe in.
Shopper-AI Trust And Inexpensive Products
Consumers worry that AI is going to give them bad recommendations and push them to spend too much money.
My company surveyed 1,463 U.S. online shoppers for the 2026 Trust in AI Commerce Report to find the maximum amount they would spend on an AI recommendation without doing their own research.
The most common answer was under $25, which was chosen by 42% of shoppers. Six in 10 capped it at $50 or less, and fewer than one in 20 will trust AI with purchases of more than $500.
In other words, the average person might be happy to rely on AI for small purchases but not for more expensive ones.
The Autonomy Curve
If you plot all the answers described above on a two-dimensional graph, you’ll get what I’m calling the autonomy curve. It slopes down as prices rise, and it never flattens out.
This curve is consistent with years of psychological research, including landmark studies by Kahneman and Tversky. People feel the pain of losing money twice as intensely as the satisfaction of gaining it. Meanwhile, gambling gets scarier as the stakes get higher. For most people, letting an AI make purchase decisions based on a “black box” algorithm they don’t understand feels like gambling.
People also distrust AI shopping recommendations even if the AI sounds like a human and offers confident, articulate answers. Shoppers will typically assume recommendations are valid until the moment they actually have to spend money; at that point they stop and verify.
When AI Recommends Costly Products
Each product sits somewhere on the autonomy curve. Only about one in 10 shoppers would let an AI agent decide a $250 purchase. And the higher the price, the less likely consumers are to trust AI recommendations.
The autonomy curve shows up in every survey that investigates this topic. A recent Checkout.com study suggested that the average consumer would not spend more than about $223 through an AI agent.
Although financial transaction systems are advancing rapidly, consumers remain unconvinced.
For example, Visa Intelligent Commerce and Mastercard Agent Pay launched in 2025 so AI agents can pay under preset limits with tokenized credentials. But Forrester found 75% of adults who shop online in the U.S., U.K. and Canada feel uneasy about allowing an AI agent to complete a purchase and pay on its own.
“But today, most consumers aren’t willing to hand over the keys,” wrote Forrester principal analyst Lily Varon.
Shopper-AI Trust Increase With Context
I believe consumers will be willing to accept AI recommendations when they know exactly where they are coming from and how they are developed.
Research suggests this could happen if an AI’s answer was completely transparent and let you see exactly where each claim came from and when it was last checked.
Capgemini asked 12,000 consumers across 12 countries and found that two-thirds of consumers will trust a digital assistant more when it explains its recommendations, and 69% are willing to let it suggest products and deals if reasons are given.
An AI that actually shows how it arrives at its recommendations and where it’s sourcing its data could be an appealing alternative to the “black box” chatbots available today.
So if you’re the CTO or product leader whose road map assumes an agent will buy on a customer’s behalf, take a look at where you are on the autonomy curve, and if you’re asking consumers to take too much on faith. Then think about how you can give consumers more confidence in what your AI agent recommends.
To do this, you will need some kind of verification technology. And my bet is that advances in verification, and not new payment features, represent the key that will unlock agentic commerce.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?