Customer-facing agents answer questions, troubleshoot, process routine requests and sometimes handle renewals or upgrades. Results can be strong — and when they go wrong, it is your company that is responsible.
What vendors report
- HubSpot's CEO said on the company's Q1 2026 earnings call (as reported by CX Today) that its Customer Agent resolves 70% of support conversations autonomously on average, up from 20% a year earlier, across more than 9,000 customers. How "resolved" is defined was not stated in the coverage.
- Salesforce's CEO said in September 2025 that the company had reduced its support headcount from about 9,000 to about 5,000 as AI agents took on work; Salesforce said it had redeployed hundreds of employees into professional services, sales and customer success.
You are accountable for what the agent says
In Moffatt v. Air Canada (2024 BCCRT 149), a British Columbia tribunal found Air Canada liable after its website chatbot gave a customer wrong information about bereavement fares. The tribunal rejected the idea that the chatbot was responsible for its own actions: it was part of Air Canada's website, and the airline was responsible for all the information on it.
Customers want a way out
Gartner's survey of 5,728 customers found 64% would prefer companies didn't use AI for customer service, and 53% would consider switching to a competitor if they found out a company was going to. The top concern: it will be harder to reach a person.
Design rules
- Ground answers in approved policy, product and contract content; test with your hardest questions.
- Visible hand-off to a person, measured by time to human.
- No commitments (refunds, discounts, contract changes) without rules or a human.
- Disclose that the customer is talking to AI where required (and usually where not).
- Measure retention, satisfaction and complaint rates, not just deflection.