AI agents force brands to define their judgement
That moment is where brand stops being a matter of colours and tone and becomes a matter of judgement. And according to one agency chief executive, it is also where AI offers companies something more valuable than speed or scale — a forced clarity about what they actually stand for.
From generating content to making calls
Most corporate conversations about artificial intelligence still revolve around output: drafting more copy, personalising more campaigns, deflecting more service tickets. Nik Nieuwenhuijs, CEO USA of Hypersolid, argues in a piece for LBB Events that these questions, while useful, miss the larger movement. [1] The technology, in his account, is migrating from producing things on a brand’s behalf to deciding things on its behalf. As agents grow more autonomous, they will determine what to recommend, when to sell, when to hold back and how to answer a customer. Eventually they may decide what kind of experience or service to build for an individual.
The gain hiding inside that shift is not automation. It is definition. An agent making a recommendation cannot lean on the intuition of a designer or a salesperson who has absorbed the company’s habits over a decade. It needs the reasoning written down, or it will substitute whatever objective is clearest — engagement, conversion, efficiency, revenue — for the brand’s intent. The act of preparing a system to decide well forces an organisation to say what it means.
Consistency of expression becomes consistency of judgement
For years, the standard test of brand consistency was aesthetic: does this look and sound like us? Nieuwenhuijs proposes a harder question. Did the system make a choice we would make? The frame moves from how a brand speaks to how it reasons, and most brand platforms, he notes, were never built to answer the second question. They describe a company as authentic, progressive, empowering, human or customer-centric — adjectives that carry a room in a presentation and offer almost nothing when a decision has to be made.
Consider the concrete forks an agent faces. Should it push or pause? Simplify or explain? Sell something, or conclude that this particular customer needs nothing right now? And what happens when the customer’s best interest collides with a commercial target? Human staff have quietly resolved those tensions for years, bringing context and experience that never quite made it into the brand book. They know when to bend a rule, when to escalate, and when a technically compliant answer would feel entirely wrong.
The uncomfortable implication is that some brands were never as defined as their custodians believed. They functioned because people kept interpreting them. An AI agent will interpret vague instructions too — but there is no reason to assume it will fill the gaps the way a seasoned employee would. Handing more decisions to systems exposes how much of a brand lives in people’s heads rather than in its strategy. That exposure is the opportunity: the tacit knowledge can finally be made explicit, examined and encoded.
A wellness agent and the question of the unnecessary purchase
The argument becomes vivid through a single imagined case. Picture a wellness brand whose agent has permission to read a customer’s sleep data, exercise patterns, calendar and purchase history. It could assemble a personalised recovery programme and recommend products at the moment they genuinely serve the customer’s goal. Then comes the test: what happens when the customer does not need another product?
Within the same evidence and safety limits, a performance-oriented brand might help that person structure a routine around a training target. A brand built around balance might help them clear space in the calendar instead. Same technology, same data, different priorities — and both would need to recognise when a question belongs with a professional. If the wellness agent must work a product into every answer, the result is advertising wearing a sophisticated personalised interface. A credible agent may sometimes have to say the customer needs to buy nothing at all, and whether a company permits that says a great deal about what the brand actually stands for.
That is the gain the shift offers: not smarter selling, but a decision architecture in which the customer’s interest can win without the brand collapsing into incoherence. The choice to refrain becomes a designed behaviour rather than an accident of a helpful employee’s mood.
Why judgement is the hard handover
Giving an AI system your tone of voice is relatively easy. Giving it your judgement is much harder. [1]. That asymmetry explains why so many brand platforms stall at the level of expression. Vocabulary can be sampled from existing copy; judgement requires a point of view about trade-offs, and trade-offs are where strategy either exists or evaporates.
Brand should not, in his framing, determine every decision an AI system makes. Product policy, legal requirements, risk, governance and commercial objectives all carry weight. What brand must become is an input into those decisions — a clear position on the relationship the company wants with its customers. Saying “we put the customer first” settles nothing. Deciding when you will accept a smaller sale so someone can make a better choice gets closer to something a system can act on.
The organisational consequence is direct. Brand needs to be present when these systems are designed, when their objectives are set and when their behaviour is tested. Summoning marketing at the end to make the agent sound like the company leaves the most consequential brand decisions to everyone else. If a brand team talks about wellbeing while a product team measures only conversion, a warmer tone of voice will not resolve the contradiction — and an agent optimising the measured number will quietly resolve it the wrong way.
Five questions that turn strategy into executable direction
Nieuwenhuijs reduces the work to a short list of questions that any organisation deploying autonomous agents needs to answer plainly. [1] Purpose: what outcome are we trying to create for the customer? Principles: how should we choose when several options could work? Priorities: what takes precedence when customer, commercial and operational interests conflict? Permissions: what may the agent decide alone, and when must it consult the customer or a colleague? Red lines: what will we refuse to do, even if it improves conversion?
The difficulty lies not in drafting answers but in agreeing on them and ensuring they surface in the product. That requires brand, business and technology people working through real situations together, and then checking what the agent actually does — especially in the moments when no obvious right answer exists. The brand book will not disappear; visual identity, voice and design remain necessary. But an agent adapting to each customer needs something more useful underneath: a stated basis for making choices.
The gain, stated without romance
As more companies gain access to comparable AI capabilities, these choices become a larger source of differentiation. Colours, language and interface will still matter, but customers will also encounter a brand through what its agent recommends, what it prioritises and what it refuses to do. Brand consistency, once a discipline of repetition, is becoming a discipline of judgement.
That is the concrete improvement at the centre of the shift. Autonomous systems do not merely automate brand expression; they force the reasoning behind it into the open, where it can be tested, argued over and made consistent. The more independent the machine becomes, the more deliberately the human side must define itself. Companies that treat the handover as a prompt-engineering chore will get an agent that optimises whatever number sits closest to the surface. Companies that treat it as a strategy exercise will get something rarer: a brand that can explain itself, decision by decision, to every customer it serves.