The Faster AI Makes Your Agency, The More You Need People Who Can Decide

For many businesses, AI has taken a lot of tedious work off our hands. In an agency like ours, that could mean pulling apart campaign data in minutes, handling the admin that used to make everyone cry tears of boredom or helping us learn a new industry in half the time. That is real, and for me, it should be going into the creative and strategic thinking clients actually hire us for.

But speed can have a side effect you might miss. When a task that used to take a human 2 hours arrives in thirty seconds, it's tempting to treat it as finished. The faster things move, the easier it becomes to nod something through without asking whether it's any good. AI is evolving quicker than anyone can reasonably keep up with, so understanding this risk is now essential.

Our chosen model of working with clients is the good ol’ retainer. A client buys our hours, and the glorious thinking that happens inside them. So the question I care about isn't how many more posts AI lets us produce. In fact, I am usually the first to suggest less is more when it comes to posting on social. They need the posts to be sharper, better argued and more human. Fewer hours on grunt work should mean more hours on what makes a difference: understanding their audience, testing an idea before it goes out, and arguing about whether a campaign is really saying anything. If AI is used only to double output, the client is losing somewhere else.

That's why at Giraffe, we use AI only for work that doesn’t require human judgement, taste or expertise. And even the areas it does help with must still have a real person steering it and reviewing it. Not as a box-ticking rule, but because the risk with these tools isn't that we all get lazy and assume they’re correct 100% of the time. A content plan made with AI might cover every point in the brief but miss the reason the client's customers actually care. An analysis that sounds certain about a trend is built, but actually it’s based on three data points and ignores the other seventeen. 

None of that is a disaster on its own. The danger creeps in like an unwanted weed. A team that gets used to work arriving polished can slowly stop interrogating it. No one can point to the exact moment complacency arrived; it just shows up as work that's fine, month after month, until a client quietly starts wondering what they're paying for. So the skill that matters more now isn't producing; it’s human judgment, knowing what good looks like, and being willing to say "not yet" to something that already looks finished.

The uncomfortable bit for me as a founder was realising that if judgement protects quality, it can't all sit with me. In a small agency, it's natural for the big calls to run through the founder or most senior team member. As the work speeds up and the team grows, that can quickly become a bottleneck at best and a single point of failure at worst. It’s not realistic for a single person to approve every post, report and recommendation. The answer isn't more sign-off. The founder's job has to shift from approving output to protecting standards: being clear about what good looks like, then giving people authority that matches their responsibility. Asking someone to own the quality of their work can feel daunting for you, but it can be exhilarating and fulfilling for your team.

This is a lesson I learned with my co-founder. Kane and I started Giraffe together in 2012, and for a long time we decided everything jointly. It felt fair, and it felt safe, but it was also slow. Every idea waited for two people to agree, and every decision waited for two people to find time for a brief argument. Eventually (perhaps a little too eventually), we drew clearer lines. As CEO, I took ideas, growth and where the business goes next. As CFO, Kane took the policy and structure that keep it running well.

It really changed things overnight for the better. We both got faster. I could move on an idea without it needing a joint decision first. Kane could build what the business needed without me hovering. What I took from it is that there isn't just one kind of judgement worth protecting. Mine is about spotting opportunities and moving on them. His is about making sure the business can hold what we build. Both are human calls, and both matter more, not less, when everything around them speeds up.

Another test for me in my quest to ‘let go’ is hiring. It’s an area of business where the stakes are high, and you live with the outcome for years. When we recruit now, our brilliant People and Operations Manager and our Head of Social lead it. They bring expertise I don't have: one in how people work and grow here, the other in what excellent social media work looks like day to day. I join the final discussion and give my opinion based on their findings. But the decision is ultimately theirs. Trusting someone else's call on something that important is harder than making it yourself. But it’s also the only way a business stops depending on one person's opinion.

AI will inevitably keep making the work faster. If you’ve been awake in the last twelve months, you’ll know that isn’t up for debate. What is up for debate is who decides whether the work is any good. A founder's job now is to make sure there's a clear answer to that, and that the answer isn't always them. Give people in your business the room to build judgment and the authority to use it. It doesn’t matter how quickly AI lets you move; deciding whether what it does is good will always be a human job. 

‍

‍

Author Bio

Phil Treagus-Evans is co-founder and CEO of Giraffe Social, a UK social media agency that helps established brands build trust on social. He is also the bestselling author of Human-First Marketing: The Art of Being Seen, Trusted and Remembered.

‍