AI without the slop: Building marketing content systems that sound like you
Jovana Lazarevska
There is a very specific feeling you get when you read a piece of AI-generated content and know immediately that it is AI-generated. The em dashes. The phrase “it is not just X, it is Y.” The oddly polished sentences that say nothing in particular about anyone in particular.
That feeling is what we set out to fix in our 10th webinar. Phil was joined by Nathan Nichols from Ikigai to get into the real problem with how most businesses are using AI for marketing content, and what a properly built content system actually looks like.
Watch the full recording here:
What is AI slop and why does it matter?
AI slop is content generated at volume with no real point of view. It is polished on the surface but hollow underneath. You could swap the brand name out and it would read exactly the same for any other business. The problem is that both humans and algorithms are getting better at spotting it.
ClickUp is a useful cautionary tale. They went from 1.2 million monthly organic visits in January 2025 to just 29,000 by April 2026, after flooding their blog with thousands of AI-generated self-promotional posts. Google eventually caught up. It always does.
The 80% stat that tells only half the story
According to HubSpot, 80% of marketers are already using AI to create content. But only 4% are using it to write entire pieces from scratch. The smart ones are using it to draft and then doing the last mile themselves. They are not outsourcing their voice to AI. They are using AI as a starting point and then making it theirs.
That distinction matters more than most people realise.
What a content system actually is
Nathan’s core idea is that the problem is not AI itself, it is how people are using it. Most people open a chat window, type a prompt, and hope for the best. What you get is inconsistent, generic, and forgettable.
A content system flips that. Instead of prompting from scratch every time, you build a set of files, rules, and checks that give the AI everything it needs to write in your brand voice. Nathan calls this the shift from prompt engineering to context engineering.
The foundation of any content system is what he calls Brand DNA: your tone of voice, what you will and will not say, your claims register, your target audience personas, and examples of content that has worked for you in the past. Get that documented properly in plain text files and the AI has something real to work with.
The live demo: from generic to on-brand
Nathan ran a live demo using Claude Cowork, building a piece of Jersey tourism content from scratch. Without the content system loaded, the output was decent but generic. Phrases like “hidden gem” crept in. The structure felt like a template.
With the content system applied, the AI caught its own mistakes, flagged banned words, removed invented facts, and rewrote the piece with specific local references. It is not perfect out of the box and Nathan was clear about that. But it is a much better starting point, and it improves each time you give it feedback.
What to do next
If you want to start building your own content system, Nathan has put together a free toolkit with the folder structure and foundational files to get you going. It is intentionally simple so that you can adapt it to your own brand.
The honest advice from both Phil and Nathan was the same: start small, do not wait until everything is perfect, and treat the system as something that evolves with you. Every time a new model is released, revisit it. Every time a piece of content underperforms, feed that learning back in.
And if the whole thing feels like too much to tackle alone, that is exactly what Nathan does at Ikigai and what we do at Blue Llama.