AI anxiety is rising almost everywhere. Except in support.

We surveyed 2,015 people who recently used AI customer support across the US, UK, and Canada. Views of AI support improved close to 2 to 1 over the past year, while general AI sentiment kept falling. Support is the exception, and this report shows why.

Why teams switch: When deflection isn't enough, Lorikeet AI takes action—updates orders, processes refunds, books appointments. Get dedicated implementation support with outcome-aligned pricing based on successful resolutions.

What you'll learn:

→ Why support sentiment is improving 2 to 1 while broad AI sentiment keeps falling
→ Which industries consumers name for good (and bad) AI support: a 2.6x vs 3.2x inverse split
→ Why 71% stop caring about human vs AI once the problem is solved
→ Why good AI support makes a brand more memorable, and earns a second chance after failure

  • Action-taking, not just answering

  • Team of specialist agents

  • Highly configurable, with PM and eng support to implement

  • Built for regulated industries

Read the report.

2,015

consumers surveyed

3

countries (US, UK, CA)

95%

recent AI-support users

Methodology: Verified Prolific panel, fielded July 2026. United States, United Kingdom, and Canada. Roughly 95% used an AI support agent in the last 3 months. No gating, no follow-up calls.

Trusted by CX leaders in fintech, health tech and complex industries

The five findings

Support is the exception to AI anxiety.

2:1

improving vs souring on AI support over the past year (38.2% more positive vs 19.3% more negative), against a falling general AI trend.

71%

would be satisfied with AI once it resolves the issue correctly and quickly. Only 9.5% still hold out for a human. The human preference is conditional on the outcome.

69%

say an excellent AI support experience makes a brand more memorable, against only 3.8% who say less. Good AI support is a brand asset in its own right.

We tested AI solutions head-to-head and Lorikeet was a winner in every metric.

Lindsay Boland, Customer Service AI Product Lead, Flex

What’s inside.

Ten sections from the report. Five findings front-loaded. Two real customer deployments. Data-led, no padding.

01

Executive summary: the five findings, front-loaded

02

Methodology: 2,015 recent AI-support users across the US, UK, and Canada

03

The support exception: sentiment improving 2 to 1 against a falling general trend

04

Quality tracks industry: Amazon, and the 2.6x vs 3.2x inverse split

05

The conditional human: 71% satisfied once AI solves the problem

06

Brand memorability: 69% say great AI support is memorable

07

The second chance: 47% retry the AI after a first-try failure

08

In practice: Flex (fintech) and Eucalyptus (digital health)

09

Key implications: five concrete plays plus the numbers that matter