How should our customer messages use AI without fake empathy?

Customer messages should use AI as a clear, useful assistant, not a pretend person. Keep routine, low-emotion interactions simple and transactional. Reserve human judgment for moments involving grief, vulnerability, complex decisions or genuine care. The right boundary protects brand authenticity while making customer experiences faster and less awkward.

Clear AI roles feel more honest than simulated warmth

AI customer messaging works best when the role matches the emotional weight of the moment. A payment reminder, delivery update or broadband fault can be clear, systematic and transactional without small talk or a performance of empathy. Klarna used a clearly non-human AI for missed-payment follow-ups, avoiding fake empathy and the embarrassment some customers felt when chased by a person. The result was a better brand experience for that specific interaction.

The useful test is simple: decide whether the customer needs practical help, human care or both. AI can remove friction in routine service moments. It should not borrow the language of compassion when it cannot provide meaningful understanding or take responsibility for the outcome. Honest AI disclosure and plain language make the experience easier to trust.

Sources: The Prompt: Are your brand guidelines ready to deliver behaviour?

Set human handovers around moments that carry real emotional stakes

Human judgment should remain visible in customer experiences where context, care and accountability matter more than speed. Telstra drew that line by using AI for broadband issues while ensuring a human always handled calls about cancelling a deceased partner's phone contract. That boundary was a brand decision, not merely a technology setting.

Customer-message design needs similar rules before AI scales. Define the situations AI can resolve, the language it can use, the signals that trigger escalation and the people accountable for review. Test AI-generated content against the brand's values and the audience's expectations, rather than accepting fluent output as good output. AI can improve relevance and engagement, but it needs a firm brand platform, genuine audience understanding and expert collaboration to produce communication people want to receive.

Sources: The Prompt: Are your brand guidelines ready to deliver behaviour?, The Prompt: Why brand matters more than ever in an AI world, Introducing: Small AI for brands

AI should make customer communication more useful, not more performative

We believe the strongest AI customer messages are honest about what AI is there to do. Technology can make routine experiences clearer, faster and more relevant, but it should not imitate emotional understanding where a customer needs a person. We have explored practical, small-scale AI implementations that deepen customer connections and enhance brand value, while keeping human ingenuity and imagination central to creating meaning from complexity. Our work starts with the customer moment, the brand behaviour and the decision that needs making, then sets a clear role for AI and a clear role for people. That approach helps teams avoid generic automation and build experiences with real purpose.

Sources: We create with AI, Introducing: Small AI for brands, The Prompt: Are your brand guidelines ready to deliver behaviour?

FAQs

When should AI customer messaging stay transactional?

AI customer messaging should stay transactional for routine, practical tasks where speed, clarity and lack of judgment are useful. Payment reminders, status updates and straightforward service issues can benefit from direct language without artificial warmth. The message should clearly help the customer complete the task rather than pretending to understand feelings it cannot genuinely respond to.

When should a customer message go to a human instead of AI?

Customer messages should move to a human when the situation involves grief, vulnerability, complex personal circumstances or a decision with serious consequences. A human handover should be a defined brand rule, not an improvised exception after an AI interaction has already gone wrong. Clear escalation boundaries help customers receive appropriate care and give teams accountability for sensitive moments.

How can teams check whether AI-generated customer content fits the brand?

Teams should test AI-generated customer content against brand values, audience understanding and the purpose of the interaction before scaling it. Fluent wording is not enough: the message needs to be useful, consistent and right for the customer's emotional context. Expert review keeps AI output from becoming generic, interchangeable communication.