Ask most people to name an AI chatbot and they'll say ChatGPT. Ask the people who've actually used a handful of these tools which one they're happiest with, and the answer changes. According to new YouGov BrandIndex data, it's Anthropic's Claude, not the market leader, that current and former users in the UK rate most positively overall.
That gap between fame and satisfaction matters for insurance. UK insurers have moved well past the "should we bother with AI" stage. Insurance Business previously reported that 55% of insurance leaders say AI is already integrated into at least some business functions, a sharp shift from the pilot-only approach that dominated as recently as 2024. With so many carriers and brokers now picking vendors for the long haul, knowing which AI brands actually keep users happy, rather than which one just has the biggest name, is useful intelligence.
YouGov's UK AI brand rankings, published in August, measure something different to the usual "which app do people prefer" survey. Satisfaction is calculated by asking only people with direct experience of a tool, current or former users, whether they're happy or unhappy with it, then netting off the two figures.
On that measure, Claude comes out on top with a net satisfaction score of 56.2. Gemini is second on 46.5, just ahead of ChatGPT on 46.0. Perplexity AI (43.7) and Alexa (42.6) round out the top five. Copilot (37.2), DeepSeek (35.1) and Apple Intelligence (32.4) make up the middle of the table, with Grok (24.6) and Siri (23.7) at the bottom.
However, in YouGov's companion piece on AI brand preference, ChatGPT is comfortably the most-preferred tool among UK consumers overall, while Claude only ranks fifth. More people default to ChatGPT. Fewer of them end up satisfied with it than Claude users do with Claude. That’s useful information for insurance brokers when deciding which model to use.
Split satisfaction between current and former users and the picture gets more useful for anyone treating AI as a long-term platform decision rather than a passing experiment.
Among people still actively using a tool, satisfaction is high across the board: DeepSeek leads on 77.2, followed by Claude on 74.5 and Perplexity AI on 70.5. ChatGPT (66.6) and Gemini (64.2) aren't far behind. Everyone looks good while people are still on board.
But it’s when users switch we see a different story. Claude retains the highest goodwill after people stop using it, at 21.8, with Perplexity AI (18.7) and Apple Intelligence (18.1) close behind. ChatGPT's former-user score drops to just 2.9. Grok's turns negative, at -4.7.
That produces some big swings between the two groups. ChatGPT's current-user score sits 63.7 points above its former-user score, one of the widest gaps in the table, alongside DeepSeek (62.5 points) and Grok (62.4 points). Apple Intelligence has the smallest drop, at 30.8 points, so people who move away from it seem to leave on better terms.
Consumer chatbot habits and enterprise software decisions aren't the same thing. But insurers don't swap core systems the way consumers swap free apps, so the size of that current-to-former drop is a reasonable proxy for how a brand's reputation holds up once the honeymoon period ends.
A consumer's satisfaction with a chatbot app isn't a perfect stand-in for how an underwriter or claims handler will feel about the same technology built into a policy admin system. But the same tension shows up in the industry's own research: enthusiasm for AI is running ahead of confidence in how it's actually used.
Insurance Business has reported that 26% of European consumers say nothing would increase their confidence in insurers' use of AI, a figure that has risen rather than fallen as adoption spreads. Separately, GlobalData polling found that close to a quarter of insurance professionals think the technology itself isn't yet ready for widespread use in the sector, citing unresolved questions over liability and regulation.
Satisfaction with AI tools is real and often high. It's also uneven across brands and easy to lose once someone has a bad experience or drifts away.
The satisfaction rankings are about consumer chatbot apps, but it's worth asking where these same AI labs actually show up inside UK insurance right now, because the picture is more mixed than "pick a favourite chatbot" suggests.
On the customer-facing side, Aviva became the first major UK insurer to launch an app on ChatGPT, letting people get an initial home insurance quote inside the chat window itself rather than on Aviva's own site. Separately, Aviva has also rolled out an internal AI tool that summarises medical reports for underwriters on life and critical illness applications, which the insurer says has cut review times roughly in half, though it hasn't disclosed which model sits behind that particular tool.
On the internal productivity side, AXA has begun rolling out Microsoft 365 Copilot to staff worldwide, a deployment that covers its UK operation and sits on top of GPT models via Microsoft's enterprise wrapper rather than a standalone chatbot. Copilot's pitch to insurers generally leans on this packaging: it's presented as an extension of the Microsoft 365 tools underwriters and brokers already use daily, rather than a separate AI product they need to learn.
Claude's footprint in UK insurance looks different again. Rather than a single named UK carrier announcing "we use Claude," Anthropic's presence has come through infrastructure and data providers that UK insurers and brokers already rely on. Analytics firm Verisk added Claude connectors for underwriting and property-restoration analytics in May 2026, and Dun & Bradstreet built commercial underwriting tools on Claude for verifying business identity and risk during submission clearance. Both suppliers count UK carriers and MGAs among their client base, so the practical effect is that a broker or underwriter could be interacting with Claude through a familiar third-party tool without the model ever being branded front and centre.
That's arguably the more realistic near-term pattern across the market: fewer insurers picking a single "house" chatbot brand, and more of them ending up with several models embedded inside the software they already use, chosen by their vendors rather than by the insurer directly.
This isn't a case for switching to whichever chatbot tops a consumer satisfaction table. Enterprise procurement rightly weighs data security, regulatory compliance and system integration far more heavily than general-population sentiment. But the YouGov findings are a useful check on assuming market share and user satisfaction are the same thing. The brand people reach for first isn't necessarily the one that leaves them most content afterwards.
UK insurers are still working through what Insurance Business has called an "operational divide" between AI ambition and day-to-day reality. Independent brand data like this is likely to feed into procurement conversations more often from here, alongside the harder questions on governance and accountability the sector is still working through.
Data source: YouGov BrandIndex, based on UK data collected between 1 March and 31 July 2026. Satisfaction is a net score derived from current and former users who say they are satisfied or dissatisfied with a given AI tool; it is not a percentage.
