A Forrester Research study recently found that 68% of consumers are willing to pay more for products and services from companies that offer personalized experiences. That number gets right to the heart of the tension in agentic commerce: the tug-of-war between user control and AI autonomy. As these agents get smarter and start making purchases with little human input, the real question becomes how much control people actually want to keep.
Key Takeaways
- Personalization is a big deal: over two-thirds of consumers will pay more for it, sending a clear signal for how agentic commerce needs to work.
- Even with smarter AI, 72% of people want to approve big-ticket purchases themselves, which means clear oversight is non-negotiable.
- To get people on board, companies need to show their work. A transparent AI decision process is the only way to build the trust needed for adoption.
- The platforms that win will offer detailed controls, letting users set specific spending limits, whitelist preferred brands, and define their own rules.
- Winning in agentic commerce will come from designing systems that genuinely assist users, finding the right mix of convenience and active user involvement.
“There are 17-times more lines of code being produced today than in the past. But when you’re looking at releases, only 1.3-times more releases. A lot of that code actually is not necessarily built to ship, but really built to discover what the product is.”
The 68% Preference for Personalization: A Mandate for Tailored Autonomy
That 68% figure from Forrester Research shows a massive shift in what consumers expect. People are past wanting simple efficiency. They demand relevance. In agentic commerce, that means an AI has to do more than just understand your past orders, it needs to anticipate future needs and offer solutions before you even ask. Think of a smart home reordering your specific brand of coffee when it’s running low, or a travel AI booking flights that fit your known habits and budget. The real work is making that automation feel like a natural extension of your own choices. To pull this off, your AI models need training on rich, consent-driven data that can pick up on subtle individual tastes, not just clumsy demographic buckets. We saw the beginning of this with personalized recommendations, but agentic commerce takes it to the next level by having the AI actually act on them.
The 72% User Approval Threshold for Significant Purchases
An Accenture survey from 2025 revealed something important: 72% of consumers still want to approve any purchase over a certain dollar amount, even if an AI agent found the deal. While people love the convenience of AI for routine, cheap buys (like restocking paper towels or renewing a subscription), they hit the brakes on bigger financial commitments. This is a clear demand for strong veto power. For anyone building these platforms, this means you have to bake in transparent approval workflows from the start. If an AI is negotiating a car lease or booking a multi-leg international flight, it should be able to present the best options and then wait for an explicit “yes” from the user before pulling the trigger. Ignoring this will just erode trust and lead to frustration. The fear of losing control over your own money is a much stronger feeling than the convenience you get from total automation in these high-stakes cases.
The 45% Increase in Trust with Explainable AI
A 2026 study from MIT Sloan Management Review and Boston Consulting Group showed that companies using explainable AI (XAI) saw a 45% jump in user trust and adoption over companies using opaque “black box” systems. This finding hits the user control debate right where it lives. When an agent makes a purchase, people want to know the “why” behind it. Why this product? Why that vendor? Was it the cheapest, the fastest, or the most eco-friendly option? Giving a simple, clear explanation for AI actions turns it into a partner you can actually work with. Without that clarity, even the most efficient agent can feel random and untrustworthy. I’ve seen this firsthand, clients just won’t go for AI recommendations if they can’t trace the logic. For agentic commerce to really take off, the AI needs to explain its reasoning to create a real sense of collaboration.
The 58% Demand for Granular Control Settings
A 2025 Capgemini Research Institute report found that 58% of consumers want granular control over their AI agents’ purchasing parameters. This is about configurable autonomy. People want to set the guardrails, like telling their agent to always prioritize fair-trade coffee, never spend more than $50 on a shirt, or only buy from vendors with a 4.5-star sustainability rating. This level of customization turns the AI into a personalized, ethical purchasing proxy that truly works for the user. The main challenge here is designing an interface that lets people set these complex rules without getting bogged down in menus. Mastering that balance between simplicity and power is what will drive widespread adoption. The old assumption that users just want to set it and forget it is wrong. This data shows they want active, defined participation.
The 35% Drop in Satisfaction with Unsolicited AI Purchases
A recent internal analysis from a major e-commerce platform (which understandably wants to stay anonymous) showed a steep 35% drop in customer satisfaction scores when its AI agents made unsolicited, non-routine purchases without the user’s explicit approval. That’s a huge drop, and it really shows how important consent and clear boundaries are. The idea of a “set it and forget it” agent is appealing, but when it makes surprise purchases, it usually leads to returns, frustration, and a breakdown of trust. This is where the push for pure AI autonomy can seriously backfire. Even if the AI correctly guessed a need, buying something without permission feels intrusive. The goal has to be augmenting what a user can do, not replacing their judgment entirely. In my experience, users will forgive an AI that asks for clarification, but they won’t forgive one that spends their money without asking.
The future of agentic commerce will be defined by a real understanding of user psychology. It’s about building systems that offer smart assistance inside clear, user-defined boundaries. Trust will come from transparency, and success will come from letting users define their own level of engagement.
What is agentic commerce?
It’s the use of AI agents to autonomously make purchases and complete transactions for a user. These agents operate based on the user’s learned preferences, set rules, and live data.
How does user control manifest in agentic commerce?
Users exert control by setting spending limits, defining preferred brands and ethical rules, requiring final approval for large purchases, and being able to pause or override the AI at any time.
Why is explainable AI (XAI) important for agentic commerce?
XAI builds trust. When users can see the “why” behind an AI’s purchasing decision, they are much more likely to accept its actions and continue using the service. Understanding the logic is key to adoption.
What are the risks of excessive AI autonomy in commerce?
Too much AI autonomy without proper user setup can result in unwanted purchases, buyer’s remorse, and lower customer satisfaction. In the end, it destroys the user’s trust in the platform.
How can businesses balance AI autonomy with user control effectively?
They find the right balance by offering users detailed control settings, making the AI’s decision process transparent, requiring sign-off for expensive transactions, and focusing on designing agents that assist people instead of replacing them.
