This article went through the process I’m about to describe. That felt like a useful starting point, because the best way to explain how AI fits into content creation is to show it rather than theorise about it. What follows is how I capture, develop, and publish ideas for BadCompost, and where AI helps along the way.
The problem worth solving
I suspect many SME leaders, and also team members, employees, anyone really, have things worth saying. Hard-won insights, perspectives shaped by experience, observations about how work actually gets done. But there’s friction between having something to say and actually saying it. Time, focus, the gap between a rough idea and something publishable. The thought sits in your head, maybe in a voice note, maybe in a half-finished document, and it just stays there.
The easy answer is to let AI write it for you. But the output tends to be generic, obvious, soulless. It reads like what it is… content generated to fill a gap rather than to say something. That’s not what I was looking for.
The other option is to do everything yourself. Which is fine if you have the time and the process, but most people don’t. The friction wins.
I wanted a middle ground. Something that amplifies rather than replaces. AI as a thinking partner, not a ghostwriter who doesn’t know you.
How the process works
Capturing ideas
It starts with getting things out of my head without worrying about structure or polish. Voice notes, brain dumps, rough conversations. The goal is volume and honesty. I’m not trying to write at this stage, I’m trying to capture what I actually think before I lose it or start editing myself.
Most of this is messy. Half-formed thoughts, tangents, contradictions. That’s fine. The raw material doesn’t need to be good, it just needs to exist.
Synthesising and shaping
This is where AI starts to help. I feed in the rough thinking and ask it to pull out themes, identify what’s worth developing, highlight where there’s an actual argument versus where I’m just circling. It’s a back-and-forth. AI proposes, I direct, we iterate.
What I’ve found is that AI is useful for seeing patterns I’m too close to notice. It can reflect back what I’ve said in a way that helps me decide what I actually want to focus on. It’s not generating ideas for me, it’s helping me see the ones I already have.
Structuring and drafting
Once I know what I’m trying to say, we work on structure together. What’s the arc? What needs to come first? Where’s the natural endpoint? AI can propose outlines, but the shape of the argument remains mine. I know what I want the reader to take away, AI helps me figure out how to get there.
Drafting is collaborative too. AI produces sections based on our agreed structure, I review and redirect. Some of it lands, some of it needs reworking. The first pass is rarely right, but it’s a starting point that’s faster than staring at a blank page.
There’s a practical side to this as well. It’s all well and good having an angle, saying something insightful, but if nobody ever reads it then it’s kind of lost. AI is also good at grounding my content in the reality of SEO, keywords, concepts that I wouldn’t necessarily think about or really care about, but that help the work actually reach people.
Refining voice and tone
This is where the real work is. AI can produce competent prose, but competent prose isn’t the same as writing that sounds like you. Early on, the output felt generic. Too polished, too smooth, missing the rough edges that make writing feel human.
Getting this right took time. I had to teach the AI what to avoid, what patterns to break, how I actually construct sentences. It’s similar to onboarding a colleague who writes on your behalf. At first you’re correcting a lot. Over time they learn how you think and what you care about, and the steering becomes lighter. AI works the same way. The more context it has, the closer it gets.
What I’ve learned
The infrastructure takes some setup. You need to think about how you capture ideas, how you feed them into AI, what prompts and guidance produce useful output. There’s a learning curve.
But the barrier to doing something is actually quite low. You can get started quickly, produce something real, and refine from there. The refinement is ongoing, just like any collaboration. You get better at steering, and the AI gets better at responding.
The important thing is that the output is still yours. The ideas come from you. The argument comes from you. The voice, with enough care, comes from you. AI is the tool, not the author.
The invitation
There are voices we’re not hearing. SME leaders with insights worth sharing who never get around to sharing them. Not because they have nothing to say, but because the friction is too high.
If that sounds familiar, this is one way in. AI won’t write your content for you… not well anyway. But it can help you get your thinking out of your head and into a shape worth publishing. The middle ground exists, and it’s more accessible than it looks.
If you’re curious how this might work for you, or just want to talk it through, reach out. Sometimes the first step is just a conversation.