Sian Conway-Wood on why AI companies are turning to cute mascots

AI companies are adopting cartoon mascots to make powerful technology feel less threatening to the mainstream users they now need, according to Sian Conway-Wood, senior managing partner at 181 Street.

Meta’s mascot is Jolly, a cheery figure with rosy cheeks. Anthropic’s is Clawd, a crab in pixel form reminiscent of a 1990s computer game. OpenAI has introduced dots, a set of cartoon AI agents that includes a smiling frog, a heart wearing sunglasses and a blue cloud in a beret.

“It’s essentially the return of the Tamagotchi, except this one might take your job,” Conway-Wood writes.

She describes the AI cuteness offensive as selling a world-changing technology with “something you’d win at a fairground”, and adds that anyone who found Labubus sinister should see what Silicon Valley has been breeding. She notes the paradox that as the technology grows more powerful, its branding looks less sophisticated. After years of persuading people that AI is extraordinarily capable, the industry now seems keen for them to forget, and to evoke instead childhood wonder. “And when the strategy starts to seem counterintuitive, I smell a cult,” she writes.

The compatibility problem

For early adopters, using AI was part of the appeal because it signalled curiosity, competence and technical fluency. But Conway-Wood argues that, as is true of every cult brand, those who drive early success are not necessarily those who deliver mass-market success.

She refers to sociologist Everett Rogers’ Diffusion of Innovations theory, introduced in 1962, which sets out how new ideas move through society, from innovators and early adopters through to a more cautious majority. Each group approaches innovation differently, and compatibility with people’s existing values, experiences and needs is a major factor in adoption.

“Being better isn’t enough. It also has to feel like something for people like you,” she writes. She links this to self-congruity theory, which explains why people are drawn to brands that reflect how they see themselves or want to be seen.

Winning early adopters meant convincing them AI was very smart. “You know this cohort. They are on X, talking about IQ and trying to convince us we no longer need artists to create art,” she writes. The next group, she argues, may resist using something presented as cleverer than they are. A tool that makes people more capable supports their sense of competence; a machine that apparently does their job better threatens it, especially for those whose identity rests on expertise, creativity or professional judgement.

She points to the reaction to the Society of Authors’ Charter of Ethical AI Use. Launched on 14 October to establish rules on transparency and responsible use of AI in publishing, it was dropped by the Society a week later after members objected.

“AI has an identity problem,” she writes, suggesting that presenting the technology as a rather helpless cartoon figure could be a clever way of addressing part of it.

Lessons from the iPod

Conway-Wood compares the situation to Apple’s iPod. Launched in 2001 on the promise of a thousand songs in your pocket, it was promoted from 2003 with silhouette ads showing people dancing against bright backgrounds, connected to their iPods by white headphones.

The technology was still impressive, but Apple no longer made cleverness the message. Its white headphones turned into a visible signal of identity that made adoption observable, and the iPod came to belong to a world of music, fashion and self-expression. “When the people you identify with start using something, it becomes considerably easier to imagine using it yourself,” she writes.

AI companies face a similar task, she says, though their visual language points to a different identity problem.

Shifting the power balance

Anthropomorphism is a long-established marketing device, she notes, citing the meerkat in a velvet smoking jacket that has sold insurance for years. It works because people readily form impressions of things that seem to have intentions, emotions and personalities. Research on anthropomorphic chatbots suggests humanlike features can increase trust and emotional connection, though the effect depends on user and context.

“But there’s something more sinister happening with AI,” she writes. The characters, she argues, may alter the perceived power relationship. A technology discussed in terms of the jobs it could replace and the decisions it might make for people becomes, when represented by a small creature with huge eyes, something that appears to need help crossing the road. “The technology hasn’t become any less powerful,” she writes. “But the relationship we’re invited to imagine having with it has changed.”

Different mascots, different users

Conway-Wood says cults are skilled at telling people what they want to hear, promising lonely people belonging, anxious people certainty and those wanting to feel special a sense of superiority. AI means different things to different people: possibility to the technically confident, saved time to a busy parent, a threat to a creative professional’s skills and another tiring thing to learn for someone who avoids new technology. These are differences of identity and values rather than demographics, she argues.

That makes the contrast between Clawd and Jolly significant. Clawd’s pixelated oddness suits developer culture, drawing on retro computing and internet in-jokes. Jolly is closer to mainstream consumer mascots, soft, friendly and uncomplicated. One works as a badge of belonging for confident users; the other seems built to make the unfamiliar feel familiar.

From first use to habit

Deloitte’s 2026 UK Digital Consumer Trends research found 63% of consumers had tried generative AI but only 10% used a standalone generative AI app daily. Conway-Wood says the figures do not pin down AI’s position on the adoption curve, but suggest a large gap between experimenting and habitual use.

She cites Duolingo, whose product relies on streaks, points, rewards and reminders. Its green owl, Duo, became the face of that system, with a needy, persistent and sometimes unhinged personality that mirrors the behaviour the app wants: coming back tomorrow. The mascot did not create the habit, she argues, but gave the habit-forming mechanics a distinctive personality that spread into advertising, social media and popular culture.

For AI companies the task is harder. A friendly mascot may get someone to ask a first question, but she questions whether it will bring them back, show where the product fits their routines or give them a reason to pick one assistant over another. If every AI company has its own cuddly creature, cuteness stops being distinctive, and the advantage goes to whoever turns familiarity into something people would miss.

The risk of looking harmless

Conway-Wood also warns that giving technology human traits can make it seem more trustworthy while doing nothing for its reliability. A smiling character does not make answers more accurate or guarantee responsible handling of data, and something that looks like a toy can still carry huge social and economic consequences.

“Making technology approachable is valuable,” she writes. “Disguising the reasons people might reasonably be cautious is another matter.” The more effectively a brand encourages people to treat its technology as if it were human, the more those distinctions matter, she adds.

She argues AI’s next phase of growth depends on understanding the identities people bring to it, giving it a meaningful place in their lives and offering enough value to bring them back. Marketers, she says, must ask whether making something powerful look harmless helps people make better decisions “or simply making it easier for them to stop asking questions”.