The Honest Answer Is "It Depends"
Ask someone if they like AI agents and you'll get a visceral reaction. Usually negative. That's because most people's experience with AI agents involves getting trapped in a loop of "I didn't understand that. Can you rephrase?" while desperately searching for a "talk to a human" button that doesn't exist.
But the survey data tells a more nuanced story. Salesforce found that 69% of consumers prefer AI agents for quick communication with brands. A Tidio study put it at 62% who would rather interact with an agent than wait 15 minutes for a human agent. And HubSpot's research found that 40% of consumers don't care whether they're helped by an AI or a human, as long as their question gets answered.
So customers don't hate AI agents. They hate bad AI agents. And there's a meaningful difference.
What Makes an Agent Infuriating
I've studied hundreds of agent interactions across different platforms, and the frustration sources are remarkably consistent. They boil down to four patterns:
1. The Infinite Loop
The agent doesn't understand the question, offers the same three options, the customer rephrases, and gets the same three options again. There's no exit. No escalation. Just a digital hamster wheel.
This is the cardinal sin of agent design. Every agent needs an escape hatch. After two failed attempts to answer, it should proactively offer a human agent, an email fallback, or at minimum acknowledge that it can't help with this specific question.
2. The Generic Non-Answer
Customer: "Why was I charged $49.99 on March 3rd?" Agent: "For information about billing, please visit our billing FAQ page."
This is technically a response. It's also completely useless. The customer didn't ask for a FAQ link. They asked about a specific charge. When an agent can't handle specificity, it feels like talking to a brick wall with a search engine bolted on.
3. The Identity Crisis
Nothing tanks trust faster than an agent pretending to be human. Customers can tell. The uncanny valley of "Hi there! I'm Sarah and I'm SO excited to help you today!" from an agent is worse than just being upfront. Transparency about being AI actually increases satisfaction scores, according to a 2024 study from the University of Gothenburg.
4. The Context Amnesia
Customer explains their problem in detail. Agent asks a clarifying question. Customer answers. Agent asks the same clarifying question again, or starts the conversation over. Every piece of context that gets lost is a piece of the customer's patience.
What Makes an Agent Great
The flip side of frustration is delight, and the AI agents that customers genuinely appreciate share specific characteristics:
Speed Above All
The number one reason customers prefer AI agents is speed. Not personality. Not cleverness. Speed. When someone has a straightforward question, getting an accurate answer in 5 seconds beats waiting 3 minutes for a human agent, every single time. Zendesk's data shows that first response time is the strongest predictor of customer satisfaction, and no human can match an AI's response time on known queries.
Accuracy That Earns Trust
An agent that says "I don't know, let me connect you with someone who does" builds more trust than one that confidently gives wrong information. Modern RAG-based AI agents solve the accuracy problem by pulling answers from verified documentation rather than generating responses from general training data. This is a fundamental shift from the AI agents of even two years ago.
Remembering the Conversation
Good AI agents maintain context throughout the conversation. If a customer mentions they're on the Pro plan in their first message, the agent should factor that into every subsequent answer without asking again. This seems basic, but a surprising number of agent implementations reset context with every exchange.
Always Available, No Guilt
There's a psychological component to agent preference that doesn't show up in surveys. With a human agent, customers feel pressure to be concise, polite, and not "waste their time." With an agent, that social pressure evaporates. People ask more questions, explore more options, and take their time. The result is often a better-informed customer who makes a more confident purchase decision.
The Generational Split
Age matters in agent preference, but perhaps not in the direction you'd expect.
- Gen Z (18-26): 73% prefer AI agents or self-service over calling or emailing. They've grown up texting and they're comfortable with asynchronous, text-based communication.
- Millennials (27-42): 65% prefer AI agents for routine queries, but want human escalation available. They'll use whatever's fastest.
- Gen X (43-58): 52% are comfortable with AI agents, but skeptical. They need to be won over with competence. One bad interaction and they'll demand a phone number.
- Boomers (59+): 35% prefer AI agents, but this number is growing as AI quality improves. The key insight: they don't dislike the technology, they dislike poor implementations.
The trend across all demographics is upward. As AI agents get genuinely better at understanding and responding, resistance decreases. The AI agents people hated in 2022 are not the AI agents available in 2026.
How Page-Aware AI Changes the Equation
Here's where things get interesting. Traditional AI agents operate blind. They don't know what page the customer is on, what they're looking at, or what they've already tried. The customer has to explain everything from scratch.
Page-aware AI, the kind we built into hiroi, fundamentally changes the interaction quality. When the agent can see the same page the customer sees, conversations get dramatically shorter and more useful.
Instead of: - Customer: "I can't figure out how to change my subscription." - Agent: "Which page are you on?" - Customer: "The settings page." - Agent: "Which settings page? Account settings or billing settings?" - Customer: "Billing." - Agent: "Ok, click on 'Subscription' in the left sidebar."
You get: - Customer: "I can't figure out how to change my subscription." - Agent: "I can see you're on the Billing Settings page. Click the 'Change Plan' button next to your current Pro plan in the Subscription section. Want me to highlight it for you?"
One exchange instead of five. The customer feels understood. The AI feels competent. This is what moves agent satisfaction from "tolerable" to "genuinely preferred."
hiroi's page integration goes further by supporting highlighting, scrolling to specific content, and even suggesting edits to form fields. The agent doesn't just tell you what to do; it can show you where to look and walk you through it visually. That's a level of assistance that most human agents can't provide over a text chat channel.
What the Data Says About the Future
Customer expectations are shifting faster than most companies realize. A McKinsey study found that 75% of customers expect service within 5 minutes of making contact online. Not 5 minutes during business hours. Five minutes at any time. That expectation is physically impossible to meet with human-only support at most price points.
The companies with the highest customer satisfaction scores in 2025 and 2026 aren't the ones with the biggest support teams. They're the ones with the smartest AI front line and the most seamless escalation to humans when needed.
The question isn't whether customers like AI agents. The question is whether your agent is good enough to earn their preference. With the right knowledge base, contextual awareness, and graceful escalation, the answer is increasingly yes.
Building an Agent Customers Won't Hate
If you take one thing from this post, make it this: customers judge AI agents on outcomes, not on the fact that they're AI. Solve the problem fast and accurately, and they'll prefer the agent. Fail, and they'll remember the frustration.
The practical checklist:
- Be transparent about being AI. It builds trust.
- Answer from verified sources. RAG over general knowledge, always.
- Know when to escalate. Two failed attempts should trigger a human option.
- Remember context. Never make customers repeat themselves.
- Use page awareness. The less the customer has to explain, the better.
- Be fast. This is your biggest advantage. Don't squander it with slow API calls or unnecessary loading screens.
The bar isn't perfection. It's being clearly better than the alternative of waiting in a queue. That bar is achievable today, and the tools to get there are more accessible than they've ever been.