stories

Something big is about to change in how people connect with brands. Until now, customers have visited your website, app, or store to obtain what they needed. Soon, many won’t bother. They’ll simply ask an AI, and it will do the work for them—finding answers, making purchases, even negotiating deals.

This isn’t sci-fi. LLMs hold conversations that feel human. They've quietly blown past the Turing test, the litmus test for artificial minds, and we barely noticed. The implications are huge: customer relationships are starting to be mediated by machines.

Too many companies still see AI mainly as a way to cut costs—replacing business processes with automation. From the customer’s side, though, the shift looks very different. Millions already use ChatGPT to search, compare, and decide what to buy. Today those consumer agents don’t negotiate directly with brands, but that moment is coming fast.

When they do, the first layer of competition will be price, and agents will predictably push every brand down toward the lowest possible deal. But if that’s all your brand offers, it risks becoming invisible. What will matter is how your brand’s qualities—its cultural relevance, its distinctiveness, its trustworthiness—are built into the way your agent communicates. Because even if machines don’t “feel” those values, they will surface them in their recommendations to people. The challenge is no longer just being findable on Google. It’s being recommended by these agents that are quickly becoming the gateway to every customer choice—not just surfaced in a list, but actively suggested as the best option, with reasoning for why your brand fits the need. You can’t buy your way into that position, at least not yet. You have to earn it by making your brand’s qualities clear enough for an agent to recognize and carry forward.

From Screens to Ambient Interfaces

Websites and apps gave us clarity and control, but only by forcing people through long click-paths to do simple things. And “personalization” never lived up to its promise. On one side, it was marketing fluff: swapping a banner here, a headline there. On the other, it was engagement bait: social feeds tuned to keep you scrolling.

Generative AI could break that mold. Instead of forcing everyone through the same flow, conversational interfaces let people just say what they want while an agent does the work in the background. Ten clicks become one exchange.

This is the start of polymorphic experiences. By polymorphic, we mean interfaces that can change form while staying true to the same underlying identity. They won’t disappear, but they won’t be static either: they will shape-shift to fit the moment, adjusting function, tone, layout, or sequence to match the need. And when no screen is present, the same brand experience can surface as a voice in your ear, guidance in your glasses, orchestration across services. Either way, the interaction flexes, stretching user experience far beyond fixed pages and rigid forms.

It Is Still All About User Experience

When we conducted ethnographic research for a bank on mortgage applications, we found that almost everyone shared the same fear of being rejected. For most, the fear was unnecessary—they would be approved—yet it was universal. The thought that lingered was simple and heavy: what will my parents think if I can’t even get a mortgage? The design challenge was not just usability, but creating an experience that eased that tension for everyone.

Insights like this become critical in a world where more interactions are mediated by agents that act like people. If we want these systems to feel human, we have to design not just for tasks, but for the hopes and fears that surround them. User research surfaces those hidden layers and helps us encode qualities like patience, empathy, and reassurance—qualities that make an interaction feel alive rather than mechanical.

To do that, brands need “experience bibles,” not just brand books. The old manuals specified color, type, and taglines; the new ones must codify behavior. How an agent greets, how it refuses, how it escalates, when it hands off, which tones are permissible and which are not—these are the patterns that make an AI feel consistent and trustworthy.

Where to Begin

The first step is modest: choose a single touchpoint and reimagine it through the lens of interaction. Don’t ask what content to display; ask what exchange to enable. Strip away the flowchart mentality and think instead about how the brand should behave in that moment. From there, codify the decisions. This isn’t about designing for every possible scenario, but about capturing the patterns that make your brand recognizable—its tone, its moments of surprise, the way it encourages or reassures, and yes, how it handles refusals or escalations. The point isn’t only to prevent breakdowns, but to create encounters that feel inventive, memorable, and unmistakably yours.

It’s also worth remembering that AI works best when it disappears. Too often it’s presented as a content factory—ads, images, or videos that draw attention to the tool itself. But in practice, AI should feel more like special effects in a film: invisible, but heightening the story. A well-designed experience shouldn’t make a customer think, “I’m talking to an AI.” It should feel natural, human, and unmistakably your brand. That’s why edge projects—playful campaigns, small utilities, cultural experiments—are the perfect place to start. They give you permission to try things, refine them, and let the AI vanish into the background of a seamless experience.

Designing for Non-Determinism

The second principle is to embrace the fact that AI does not behave like traditional software. It will not always give the same answer twice. Instead of treating that as a flaw, design for it. Test for variance, not just for accuracy. Build systems that can regenerate answers, escalate when needed, or hand off gracefully. Perfection is impossible; containment and recovery are what build trust.

This is where safety and brand integrity converge. Multi-agent pipelines, overseer processes, and human-in-the-loop checks aren’t technical luxuries—they’re brand necessities. They ensure that even when an interaction drifts, it stays within bounds and retains the right voice. A system that fails safely will always outperform one that promises flawless control and breaks when it falters.

The Long Game

Over time, these edge experiments become more than curiosities. Each attempt adds to a repertoire of behaviors that form a living signature. As you build this repertoire, it shifts from the margins to the center of your brand. What began as experimentation becomes the backbone of your experience, a system robust enough to guide every interaction at scale.

But it’s important to remember that AI is not the brand. It doesn’t invent your culture, your story, or the meaning of your product. Those come from people—the creative choices, the craft, the values that make your brand worth caring about in the first place. AI’s role is to make those qualities felt in every interaction. Done well, it disappears, leaving only the story and the emotion.

This is the long game. AI won’t reward hesitation. The brands that thrive will use it as a stage for their personality, not a substitute for it—starting with small, invisible experiments at the edges and carrying those patterns to the center once proven. That’s how you keep the human core intact while ensuring your brand feels alive when everything else flattens into parity.

About The Author

Jean-François Lavigne is Creative Director & Innovation Lead at Sid Lee. He works at the intersection of interaction design, branding, and AI, helping turn new tools into meaningful products and brand experiences.