What did the Capacity survey find?
On October 27, 2026, Capacity published The Agentic Consumer Report, a survey of 1,000 US consumers. It found that 35% have already sent their own AI agent to handle a customer service request, and 44% say an AI tool has contradicted a company representative or the company's own policy.
The other numbers in the release on PR Newswire:
- 55% have used some AI tool, an agent or a chatbot like ChatGPT, to help with a customer service request.
- Among Gen Z and Millennials, 49% have used a personal AI agent for customer service.
- 34% say AI makes them ask more specific questions when they deal with a company.
- 23% would ask a representative to explain or cite company policy because of something AI told them.
- When AI and a company representative disagree, consumers are about equally likely to believe either one.
Capacity sells customer service automation, so treat this as vendor research. The release doesn't give field dates or name the panel provider, a gap The Agile Brand Guide also points out. I still think the direction is right, because I hear it on our own calls.
Why does it matter that 44% say AI contradicted a company?
A big share of your customers now show up with a second opinion already in hand, and that second opinion is often wrong about you. If ChatGPT told a lead your service costs $99 or that you offer free cancellations, your rep is now arguing with a source the customer trusts as much as the rep.
The AI usually isn't lying on purpose. It fills gaps. If your website doesn't say what you charge, how refunds work or which areas you serve, the model guesses from reviews, competitors and old pages. Then the customer calls you quoting the guess.
Capacity's founder and CEO, David Karandish, said in the release that businesses "need to rethink how they measure success, moving beyond speed and efficiency." I agree with half of that. Speed still matters a lot. A fast answer now also has to match what every other source says about you.
What does this mean if you run ads, answer leads or close sales?
Your lead may have researched you with AI before the first call, and some will have an AI agent message or call you on their behalf. Whoever answers, a person or your own AI, has to give the same answer every time, quickly, and from information you control.
Here's where it shows up:
| Where | What changes | What to do |
|---|---|---|
| Ads | People click, then ask ChatGPT about you before they fill out the form. | Put pricing ranges, service areas and guarantees on the landing page in plain words. |
| Answering leads | Some first messages will come from a customer's AI agent, with very specific questions. | Answer within minutes, with the same facts your site shows. |
| Sales calls | Prospects quote things AI told them, some of it wrong. | Ask early what they've already read, then correct it calmly with your written policy. |
| Support | 23% may ask your rep to cite the policy. | Keep refund, cancellation and warranty terms on one public page and link to it. |
Speed still wins the deal. The Harvard Business Review study of lead response found that companies that tried to reach a lead within an hour were nearly seven times more likely to qualify it than companies that waited longer. An AI agent shopping for a plumber won't wait until Monday for you to text back. I wrote more about that in why leads go cold.
We built the KenjiAI agents to answer from your own business info for this reason. Conversation AI handles texts and chat, and the Voice AI agents pick up calls and book appointments. They reply in seconds, day or night, with the same prices and policies your team uses, so a customer's AI gets the right answer the first time. We've run ads since 2013 for 500+ businesses, and the leads that convert are the ones that get a fast, consistent reply.
What should you fix this week?
Write down the five facts customers ask about most, publish them on your website in plain language, and make sure your team, your texts and any AI agent you use say exactly the same thing. That page becomes what AI tools read and what your reps point to.
- Ask ChatGPT, Gemini and Perplexity what you charge, what you offer and what your refund policy is. Write down every wrong answer.
- Fix each wrong answer at the source: your website, your Google Business Profile and your service pages.
- Put one policy page live (pricing ranges, cancellations, refunds, warranty, service area) and link to it in your text replies.
- Check how fast you answer a new lead today. Text your own business from another phone after hours and time the reply.
- Have your closers ask "what have you read about us so far?" in the first two minutes, so wrong info comes out early.
Start with step 1 tonight. It takes ten minutes, and you'll see what your next lead already believes about you.