It’s 2025, and let’s be honest: the “AI Slop” epidemic is a well-documented crisis. John Oliver has masterfully roasted it, the hosts of Hard Fork have dissected it on their podcast, and we all scroll past it every single day. We know what it looks like: generic, soulless content that sounds like it was written by no one in particular.
That’s why the main headline on the DIALØGUE homepage isn’t about instant results; it’s about “Your Creative Control.”
From the very first line of code, my goal was to build a tool that respects the user’s expertise. But my breaking point came when I realized our early prototypes were failing that mission spectacularly. I’d prompt the system for a podcast about, say, marketing strategy (something I actually know about from my advertising days), and the AI would generate “natural dialogue” that was anything but. It confidently explained concepts I’d never heard of, used jargon incorrectly, and somehow managed to sound both authoritative and completely wrong at the same time. It was bland, agreeable, and lacked any point of view – basically the opposite of every good creative brief I’d ever written.
That failure forced a decision. I threw out the “fully automated” dream and rebuilt the entire workflow around a new philosophy: AI as a collaborator, not a replacement. We embraced the idea that real quality takes ~10 minutes, not 10 seconds.
This post isn’t another exposé on the AI slop problem—we all know the problem. It’s a look at the specific, 4-step workflow we designed to solve it, ensuring that you can review before you generate and that your expertise, not the AI’s, is what shines through.
The Great AI Misconception: Why Your AI Content Sounds Like Everyone Else’s
Here’s the thing about Large Language Models: they are designed to be agreeable/helpful. They are trained on the average of the entire internet, which means they are masters of the median, champions of the commonplace. They don’t have your battle-tested personal stories, your controversial opinions, or your hard-won wisdom from a decade in your field.
When you let the AI lead the creative process, you are outsourcing your most valuable asset: your unique point of view. You’re asking a machine that is designed to be average to represent your exceptional expertise. For weeks, I was stuck here. I was trying to force the AI to have a personality, to have my voice. I wrote increasingly desperate prompts like “sound casual but professional, like someone who knows what they’re talking about but isn’t pompous about it.” The AI’s interpretation? Content that somehow managed to be both pompous AND wrong. It was a frustrating, expensive failure (and yes, I’m still a little bitter about those API costs).
The Shift: A 4-Step Workflow for AI-Assisted Expertise
The breakthrough wasn’t a better prompt; it was a better workflow. I realized the goal isn’t to get the AI to think for you, but to help you structure your own thinking. I stopped trying to make the AI the star of the show. The expert (that’s you!) has to be the star. The AI is the producer, the editor, the assistant.
Here is the 4-step, human-in-the-loop model I landed on.
Step 1: The Expert “Brain Dump”
Before you even open a new tab, you start with you. Open a blank document and write down your raw, unfiltered thoughts on your topic of expertise. Your opinions, your stories, your data points. It should be messy. My brain dumps are chaotic – they’re full of half-formed sentences, random parentheticals, and notes like “maybe mention that thing from Tuesday??” (I never remember what thing from Tuesday). And that’s the point. You are capturing your genius in its most authentic, unpolished form. My brain dump for DIALØGUE as an example below, see that I don’t care about misspellings 😀
Expansion for more features?
- Use generate queries as part of the pre feedback workflow?
Variety of voice, rythm and toneStep 2: AI for Structure, Not Substance
Now, you take that messy, brilliant document and give it to the AI. But your prompt isn’t “write a blog post from this.” It’s “Take these chaotic notes and organize them into a logical outline.” The AI’s job is to be a brilliant, tireless editor, finding the patterns in your thinking and giving you a clean structure to work with.
Step 3: The “De-Slop” Edit (This is the Magic)
This is the most important step. The AI gives you back a clean, logical outline. Now, you, the human expert, make it brilliant. You add back the nuance, the personality, the stories that only you can tell. You rewrite the AI’s generic phrasing to sound like you. For instance, the AI might suggest “discuss the importance of user feedback.” You change it to “that time a user’s feedback made me rebuild the entire workflow at 8pm (worth it).” The AI built the house’s frame; now you do the interior design. You choose the paint, the furniture, the art. You make it a home.
Step 4: AI for Production
Once the outline is perfected by you—once it truly reflects your expertise and voice—then you can use the AI for the final production. For DIALØGUE, this means generating the podcast audio. For a blog post, it might mean asking the AI to flesh out the sections based on your detailed, human-edited bullet points. The AI is no longer thinking creatively; it’s executing a well-defined plan that you created.
How I Built This Philosophy into DIALØGUE
This workflow was so effective for me that I made it the core of DIALØGUE. The interactive outline editor is the physical manifestation of Step 3. It’s the “de-slop” station. It’s designed to force a pause, to ensure the human expert is the final arbiter of quality before any audio is generated.
This philosophy extends to other features, too. You might notice DIALØGUE doesn’t automatically add intro/outro music. This is intentional. The final audio is clean, ready for your branding. I want to encourage a final, thoughtful review where you, the creator, add your own music and personality. It’s the last human touchpoint that makes it uniquely yours.
And what if you don’t like the AI voices? That’s perfectly fine! In fact, it’s another way to use the tool. You can take the final, perfectly structured script that DIALØGUE generates and use it as a teleprompter script to record in your own voice. The goal is a great final product, and you have complete control over how you get there.
Conclusion: Amplify, Don’t Abdicate
My goal is to build tools that amplify expertise, not abdicate it. I want to make experts more efficient without turning them into generic content machines. The future of content isn’t about who can generate the most, but who can generate the most value. And value comes from a unique, human point of view.
But this is just my journey (and honestly, I’m still figuring parts of it out). Have you struggled with ‘AI slop’? What workflows are you using? I’d love to hear your thoughts. Feel free to leave a comment or shoot me an email – especially if you’ve found a better way, because I’m always ready to steal… I mean, learn from good ideas.
And if you want to try the ‘thought partner’ workflow for yourself, you can take DIALØGUE for a spin. The outline editor is waiting for your genius. Or at least your slightly organized chaos, which is all I ever bring to it.