How to Fix AI Slop: Better Context, Stronger Brands and the Return of Human Judgement with Pacey.ai

How to Fix AI Slop: Better Context, Stronger Brands and the Return of Human Judgement

What happens when AI makes it easier to produce marketing content, but that content all starts to look and sound the same? In this episode of The Digital Marketing Podcast, Ciaran Rogers speaks to Evan Greally, founder and CTO of Pacey AI, about avoiding AI slop and getting back to the work that matters: understanding customers, developing ideas and building a distinctive brand.

Introduced by Daniel Rowles, the conversation draws on Evan’s experience as Head of Creative Tech and Innovation at Droga5 Dublin, part of Accenture Song, and more than 15 years working where creativity and technology meet. What begins with an AI-powered mint plant becomes a wider discussion about data overload, creative control and whether AI can develop something resembling taste. Along the way, Evan explains the thinking behind Pacey, with lessons that extend well beyond any single platform.

In This Episode

From a talking mint plant to a conversation with nature: Evan shares the story behind The Talking Tree, a project that combined environmental sensors, plant responses and AI-generated conversation. It is an unusual starting point with a serious creative question behind it: could technology encourage people to reconnect with the world around them, rather than spend more time staring at a screen?

Why marketing needs to ask “so what?”: Customer reviews, social comments, competitor activity and dashboards can all contain useful information. But what should a marketer actually do with it? Ciaran and Evan explore the gap between collecting data and making decisions, and why adding another tool can make that gap wider rather than close it.

The value of “honest silence”: Evan introduces Pacey’s principle of surfacing what matters without manufacturing urgency. He describes a system that looks for support from three or more sources before highlighting an issue, but can also tell a team there is nothing significant to worry about that day. The aim is to focus attention, not create another stream of notifications.

Can AI really have taste? Evan argues that AI can develop a useful form of brand-specific taste, while distinguishing it from the personal experience and judgement that shape human taste. He explains how approved copy and visuals help build a brand’s taste profile, and why evaluating work from an audience member’s perspective may be more revealing than asking a model to give it a numerical score. Would someone actually stop scrolling?

Turning a daily briefing into useful action: From a short, podcast-style morning update to content creation, scheduling and performance feedback, Evan outlines how Pacey connects market signals with marketing activity. The discussion explores what changes when a team can move from understanding an opportunity to developing a response without repeatedly copying information between disconnected tools.

Why editable creative matters: Ciaran highlights the importance of layered design files rather than finished images that are difficult to change. Evan explains how this approach supports human editing, reusable templates, a brand’s own photography and different identities across multiple outlets. He also previews an editable video workflow using customers’ own footage, rather than relying entirely on generated visuals.

An orchestra, not another isolated instrument: Could marketers build something similar by connecting their own AI tools? Evan acknowledges the potential for prototypes, but explains why filtering data, maintaining context, applying design constraints and continually evaluating results are harder to reproduce. The conversation uses an orchestra analogy to explore how different models and specialist agents can work towards the same creative goal.

Building trust before speeding things up: Evan shares a product-development lesson: giving people a login is not the same as helping them adopt a new way of working. Slower onboarding, visible sources and clear explanations of how conclusions were reached became central to building confidence. Throughout the discussion, people remain responsible for deciding what is right for the brand and approving what gets published.

Key Takeaways

Start with the marketing decision, not the dashboard. Before collecting another metric or adding another tool, ask what it will help you understand, change or prioritise.

Give AI direction, not just information. Relevant customer signals, a clear brand perspective and a defined purpose matter more than handing a model an undifferentiated pile of data.

Treat creative approval as a learning opportunity. Human choices about copy and design can help establish what good work looks like for a particular brand. That judgement should shape the process, not disappear from it.

Keep the work editable and the reasoning visible. Teams need to be able to refine creative assets and inspect the evidence behind recommendations, rather than simply accept whatever a system produces.

Use automation to make room for better thinking. The ambition discussed here is not to fill every channel with more posts. It is to give marketers more time to understand their audience, develop ideas and deliver the right message at the right moment.

Can AI learn a brand’s preferences well enough to demonstrate taste, or does that word belong to human creativity? Listen to the conversation and share your perspective with the team.

As Daniel explains in the introduction, The Digital Marketing Podcast has no commercial connection to Pacey AI.

📥 Access the show notes, tools, and links at: https://targetinternet.com/resources/how-to-fix-ai-slop