How sound, emotion and cognitive science are quietly shaping which AI tools billions of people trust — and what B2B brands can learn from it.
The most consequential branding decisions in technology right now aren't happening on packaging or billboards. They're happening in the two or three syllables companies choose to name their AI models.
ChatGPT. Claude. Gemini. Llama. Copilot.
These names each encode a psychological strategy, a bet on how the human brain will process, feel about, and ultimately trust the technology behind it. The generative AI market has surpassed $59 billion according to Statista, and the names at the top of that market are shaping adoption patterns for entire industries.
Yet most companies launching AI-powered products, features or internal tools choose names the way they choose conference room names: quickly, by committee and without much thought about the cognitive machinery they're activating. That's a problem, because naming isn't cosmetic. The psychology most brands miss is the difference between a product people try and a product people trust.
Your Brain Decides Before You Do
Before a user ever opens an AI tool, their brain has already made a judgment about it, based almost entirely on the name. That's not unique to AI of course. We judge everything, constantly. But what makes naming so consequential is the specific mechanism doing the judging.
Researchers have consistently found that when people encounter a name that's easy to pronounce and process, their brains interpret that ease as a signal of familiarity, safety and truth. Psychologists call this processing fluency, and the effects on decision-making are well documented.
Psychologist Adam Alter's research suggested a correlation between name fluency and stock market performance: companies with easier-to-pronounce names saw stronger returns immediately after their IPOs. Whether that's causation or just pattern is debatable, but the underlying principle is consistent with broader fluency research. The brain equates cognitive ease with credibility. Across multiple experiments, researchers have shown that people perceive sellers, partners, and brands as more trustworthy when their names are easy to pronounce, even when participants know the names were randomly assigned.
The shortcut is simple: if I can process it easily, it must be safe. For AI companies asking people to trust their technology with sensitive work, that dynamic isn't a footnote. It's the front door.
The Bouba-Kiki Effect: How Sounds Shape What We Feel
Processing fluency explains how easily a name is absorbed. But a separate phenomenon, the Bouba-Kiki effect, explains the emotional impression a name creates before any conscious evaluation begins.
In a now-famous experiment first conducted by Wolfgang Köhler in 1929 and replicated across cultures and languages, researchers showed people two shapes: one round and blobby, one jagged and angular. When asked which was "bouba" and which was "kiki," roughly 95% of people assigned "bouba" to the round shape and "kiki" to the angular one. This held across languages and cultures. It appears to be hardwired.
Our brains map speech sounds to physical sensations: round, open vowels like "oo" and "ah" feel warm, soft, and approachable. Sharp consonants like "k" and "t" feel precise, energetic, and cutting. Every name is broadcasting an emotional signal, whether you designed it to or not. Soft, voiced consonants and open vowels create warmth. Hard plosives and tight vowels create edge. The name lands in the listener's mind already carrying feeling before a single feature ever gets evaluated.
That's the landscape AI companies are navigating: powerful, opaque technology that needs to feel trustworthy enough for people to weave into how they work and think. And the name is often the first, and sometimes only, chance to set that emotional tone.
Five AI Names, Five Psychological Strategies
ChatGPT (and Openclaw): When Brand Architecture Breaks Down
"ChatGPT" is technically an acronym layered onto a function word: "Chat" plus "Generative Pre-trained Transformer." On paper, it shouldn't work. It's not particularly fluent. It's not emotionally warm. And yet it dominates, with over 800 million weekly active users.
The psychology is novelty capture combined with first-mover advantage. "ChatGPT" carried zero prior associations when it launched. No emotional baggage, no competing mental model. The brain encountered it as a completely new category label, and because it arrived first, it became the default reference point for an entirely new behavior. Every subsequent AI name gets processed in relation to it.
The lesson: being first can override the fluency disadvantage. But that trick only works once.
This is what makes the Openclaw story so interesting. OpenClaw is an open-source AI agent framework that lets AI actually control your computer, executing tasks across apps rather than just generating text. It was built by developer Peter Steinberger as a side project in late 2025, originally called "Clawdbot" (a nod to Anthropic's Claude). After a trademark complaint from Anthropic, Steinberger cycled through two more names in five days, landing on "OpenClaw" partly because Sam Altman confirmed it wouldn't conflict with OpenAI's branding. Then in February 2026, OpenAI acquired the project and brought Steinberger on board.
Try saying "Openclaw" out loud. The hard "k" snaps. "Claw" carries predatory weight. Grasping, aggressive, sharp. Through the Bouba-Kiki lens, it's almost entirely Kiki: hard plosives, closed vowels, imagery that grips rather than supports. In a category where the fundamental user question is "can I trust this?", the name introduces friction that ChatGPT's neutrality carefully avoided.
A solo developer naming a side project under trademark pressure in five days? Understandable. But OpenAI, a company valued at hundreds of billions, chose to bring that name into its brand ecosystem without considering how it fits. The acquisition was an acqui-hire, not a rebrand. OpenClaw kept its name, its identity, its aggressive sound profile. And now it sits alongside ChatGPT in OpenAI's portfolio, sending a completely different emotional signal.
That's a brand architecture gap. Not a creative failure from the original developer, but a process gap from the acquiring company. The psychology behind ChatGPT's neutral positioning wasn't extended to evaluate what a new acquisition would feel like sitting next to it.
Claude: Warmth by Design
"Claude" is a human name, and psychologically, that's doing a lot of work. Human names activate social cognition circuits. We process them differently than product names; they trigger expectations of personality, reciprocity and relationship.
The sound profile reinforces it. "Claude" is almost entirely Bouba: soft "cl" onset, open "aw" vowel, gentle "d" close. It feels approachable, measured, calm. The exact emotional territory Anthropic wants to occupy as the safety-focused AI company. There's also a phonetic echo worth noting: "Claude" sounds a lot like "cloud," which, intentional or not, reinforces an association with the infrastructure these AI companies are built on. The name says collaborator, not machine. It trades technical impressiveness for interpersonal trust, which for enterprise buyers evaluating AI for sensitive work, often matters more than a benchmark score.
Gemini: Integration as Identity
"Gemini," the twins, evokes plurality: many things working together at once. It's a known word and easy to pronounce, which gives it strong processing fluency. The sound profile is more mixed than it first appears though. The hard "g" onset and the sharp "eye" ending both lean Kiki, while the softer middle syllable pulls the other direction. It's not a purely warm name. But the word's familiarity compensates. People already know it, already have associations with it, and that pre-existing comfort does much of the trust-building work that sound alone might not.
The deeper strategy is ecosystem signaling. Google doesn't need Gemini to feel like a standalone personality. It needs Gemini to feel like a natural extension of tools people already use: Search, Workspace, Android. The name's familiarity and neutrality serve that goal. It doesn't demand a new relationship; it suggests it already belongs. Worth noting that some of this incubation happened before the name itself landed. Google's extended teasing of the product during its earlier development built anticipation that the "Gemini" label then inherited.
The risk is the flip side of that same coin. Familiar names can feel generic. When every product in an ecosystem gets AI features, a name that blends in too easily can fail to create the distinctiveness needed for independent brand equity. Gemini may end up being less a brand and more a feature label.
Llama: Disarming by Association
Meta named its open-source model family "LLaMA," which stands for Large Language Model Meta AI, conveniently spelled like the animal. The incongruity is doing real work here.
The soft "l" sounds and open vowels feel warm and non-threatening, which the Bouba-Kiki framework would predict. But the bigger play is the animal association itself. Llamas are perceived as gentle, slightly quirky, harmless. And there's a deeper layer for anyone who remembers Winamp's llama mascot: the animal has been a beloved symbol in developer culture for decades. For an AI model built by one of the world's most scrutinized technology companies, that combination of warmth and developer-community credibility is deliberate. It shifts the perception from "surveillance company does AI" to "friendly open-source community project."
Whether it fully succeeds is debatable. Developers know where LLaMA comes from, and Meta's reputation in that community is complicated at best. But the naming strategy at least gives the product a fighting chance to be evaluated on its own terms rather than its parent company's baggage.
Copilot: When a Metaphor Becomes a Ceiling
Microsoft went with functional metaphor. A copilot assists, supports, stays in the background. The name promises utility without autonomy, a safe framing for enterprise buyers nervous about AI replacing human judgment. It's also worth remembering that "Copilot" replaced "Cortana," Microsoft's previous AI assistant brand that never gained the traction they wanted. The rebrand was partly an attempt to signal a fresh start, but it also meant the new name had to carry extra weight: not just introducing itself, but distancing itself from a predecessor people had already written off.
The problem showed up later: overextension. Microsoft applied the Copilot name to Windows, Microsoft 365, GitHub, Edge, Bing, Azure, and dedicated hardware. When a single metaphor gets stretched across that many contexts, it stops creating meaning and starts creating noise. The processing fluency advantage, that feeling of easy to understand, collapses when the same word means different things depending on where you encounter it.
The result is brand dilution. Users report confusion about what "Copilot" actually is in any given context. The name that was chosen for clarity has become a source of the exact cognitive friction it was designed to prevent.
The Strategic Gap Most Brands Don't See
OpenAI defined this category. Its flagship name became synonymous with generative AI itself. And yet when it acquired OpenClaw, a product with aggressive, predatory sound symbolism, it brought the name straight into its portfolio without evaluating the brand fit. If that can happen at the company with the most to lose, the gap is obviously not about talent or resources. It's about process.
Most organizations don't have a naming process grounded in psychology. They have a brainstorm, a shortlist, a legal check, and a launch. The cognitive and emotional dimensions are either assumed to be obvious (they aren't) or dismissed as subjective (the research says otherwise).
And this isn't just a tech-giant problem anymore. A growing number of B2B companies are integrating AI into their products and services whether they planned to or not. They're naming AI-powered features, internal tools, client-facing platforms, and automation workflows, choosing names in sprint retrospectives and Slack threads, with no framework for understanding what those choices are doing in people's heads.
We see the fallout in three recurring patterns:
The Technical Trap. Engineering-led teams name products for what they do rather than how they should feel. Names get packed with abbreviations, version numbers, and jargon that create disfluency. The research is unambiguous: difficult-to-pronounce names are perceived as riskier and less trustworthy. Call your AI feature "IntelliSync Pro v3" and you've introduced cognitive friction at the exact moment you need cognitive ease. And when a name like "Openclaw" carries aggressive sound symbolism into a trust-dependent category, the friction goes deeper than pronunciation. It hits people at the level of instinct.
The Friendly Overcompensation. Marketers who understand the trust problem sometimes overcorrect, choosing names so soft and whimsical they fail to convey capability. A name that sounds like a children's app undermines confidence in enterprise buyers who need to justify the purchase to a CFO. When a name's emotional signal doesn't match the product experience, it creates a nagging sense that something doesn't fit. Not outright distrust, but a quiet erosion that chips away at adoption over time.
The Metaphor Ceiling. Companies pick a metaphor that works beautifully at launch ("Copilot," "Autopilot," "Navigator") then discover it can't stretch as the product evolves. The name that once clarified now constrains. It's the naming equivalent of building a brand on a single campaign idea: it works until it doesn't, and by the time you realize it, the name is load-bearing and you can't swap it out without starting over.
All three share the same root cause. Naming got treated as an isolated creative exercise, disconnected from the broader brand strategy it should serve. Nobody asked the foundational questions before the brainstorm started: What emotional response should this name create before someone uses the product? What cognitive associations will it trigger? Does the sound profile match the brand personality we're building? Will it cohere with the rest of the portfolio? Will it still work when the product outgrows its current form?
Why Brand Strategy Has to Come First
The psychology of naming isn't academic trivia. It's a strategic input that belongs at the beginning of every brand initiative, not tacked on after the positioning, messaging and architecture decisions are already locked.
At Kuno Creative, brand strategy starts with understanding the human on the other side of every touchpoint. The emotional encoding of a name, the cognitive load of a visual system, the narrative arc of a messaging framework. These are structural decisions that shape whether a brand gets trusted, adopted and remembered. They're not decoration. They're architecture.
When we work with clients on brand strategy, naming is never an isolated exercise. It's part of a process that accounts for audience psychology, competitive positioning, sound symbolism and long-term brand architecture. A name doesn't just label a product. It shapes how people feel about it before they ever use it.
The AI companies spending billions on model development understand some of this. But as the OpenClaw acquisition shows, even category leaders don't always extend that thinking to their brand architecture. The question for every brand building AI into its products and services is whether they'll treat naming as the strategic discipline it actually is.
The ones that do will build trust faster. The ones that don't will wonder why adoption stalls — and never think to look at the name.
Daniel Ulichney
