Deploying agentic AI? Involve your brand and marketing team

Last month, I wrote about how agentic AI changes the way your brand gets found. Today I want to look at what happens when you put it to work inside your own business.

Most of us use AI to answer questions. Agentic AI goes further than that. Instead of answering a prompt, it acts on your behalf. You set a goal and the machine works out how to reach it, step by step, without any involvement from your side.

A growing number of businesses are already using it to reduce costs, save time, handle volume at scale and stay competitive. Walmart runs negotiations with lower-value suppliers this way, around the clock, without human involvement. Payment service provider Klarna has their AI assistant handle two thirds of their customer service conversations, freeing their team for the interactions that need a more personal touch.

Where brand and marketing come in

Typically, the decision to adopt agentic AI sits with IT or senior leadership. Brand and marketing are often brought in once the system is already taking shape. But consider: if a system communicates with your audience and makes choices autonomously, would it impact customer experience? Would your AI apologise first or go straight to solving the challenge? Would it communicate differently with a frustrated customer than with someone asking a simple question? Just like the way you brief your customer service team, those instructions define how your brand comes across, every single time.

Klarna, the fintech company mentioned above, initially went too far with automation, at the cost of customer experience and brand perception. They decided to keep AI for routine tasks and rehired human support staff for more complex situations. As a Klarna spokesperson put it: "AI gives us speed. Talent gives us empathy."

If you are considering agentic AI, your brand and marketing teams belong in that conversation from the start. The technology may be flawless, but if the AI agent sounds wrong, feels off-brand, or struggles to connect with people, it will not reach its potential. What users actually experience is not the technology itself. It is how the system speaks to them.