In this episode, we are moving from strategy to execution. I’m going to walk you through the content generation automation I’ve built.
In our previous episode called Priorities, we identified that generating resource content for charity golf outings was our biggest automation win. Why? Because we have all the required assets and its potential impact on PlayThru (the side hustle we're AI-transforming) is significant.
But identifying the target is the easy part. Hitting it consistently—without hiring a massive team—is the challenge.
In this episode, we are moving from strategy to execution. I’m going to walk you through the content generation automation I’ve built. We will look at the "Control Center" that manages the workflow and dive into how I used AI not just to write, but to research, identify keywords, simulate subject-matter experts, and act as a ruthlessly efficient editor.
In many ways, AI automation isn’t too dissimilar to regular software. Most of the work runs in the background, but it's helpful to have a control center to manage what the automation does and when.
For content generation, my control center is an editorial calendar built in Google Sheets.
This spreadsheet isn’t just a list of topics; it is the dashboard that triggers the automation, and it’s where I can quickly access any of the assets generated by the AI automation.
How it works:
This setup ensures that the AI doesn't run wild. It only produces content when I, the human, determine I want content generated.
The last step of the automation is designed to update the editorial calendar status to “Draft Ready.” This stops the automation from writing the same piece of content again and again and again (because it will. Trust me!)
One of the main reasons I like using automation tools to help write content is that I can isolate and design functions to handle each step of the process, rather than trying to bake it all into one chat.
As you'll see below, when writing content for PlayThru, I want a series of steps taken for each draft. If I were writing the content from scratch, I'd do the same thing (which is why I want it off my plate; it takes too damn long). For example:
That’s a lot of steps, and for me—a person doesn’t write or read quickly—it takes a really long time.
So, with automation technology, I can have the integration tool take care of each of these steps, one at a time, and then compile everything into the Ai tools in a structured format to generate the content draft.
To prevent generic AI fluff, I have the workflow set up to do different types of research:
Online Research
We use Perplexity to scour the web for current trends, competitive gaps, and common themes related to the specific blog topic.
The Subject Matter Expert (SME):
Next, using a prompt, I asked AI to be a renowned expert with 20+ years of experience in running and managing charity golf outings, who also has vast experience in [INSERT BLOG TOPIC HERE]. The automation has this prompt built in, with variables that let me feed different details in each run.
I prompt the AI for talking points only. Make sure to tell it you don’t want intros, summaries, or explanatory text—just high-level, expert insights.
Keyword Strategy:
Instead of paying for expensive SEO tools (I’m cheap), I connected to Google Search Console. The automation pulls query data from the last 3 months, filtering out branded terms.
An "SEO Expert" AI prompt then analyzes our existing traffic to find relevant keywords where we already have a foothold, ensuring the new content reinforces and hopefully builds upon current rankings.
If you ask ChatGPT to "Write a blog post based on this research," the result is usually mediocre. To get high-quality output, we break the drafting process into three distinct operations.
The first part of the writing process is to draft an outline. I prompt OpenAI to write a thorough outline on the blog post topic (from the editorial calendar). But, I want the outline to include the research the automation has already collected.
So, I layer this information into the prompt using variables. In Make, this is very easy. You grab the output of a previous operation and drop it in. But I don’t just want to slap it on after the prompt instructions.
To ensure the AI knows what it has to work with, each piece of context is labeled and defined. To do this, I used Structured Prompting.
Structured prompting is the process of bracketing parts of your prompt in tags, like <online_research> … </online_research> or <keywords> … </keywords>. This approach clearly tells the AI where a specific part of the prompt starts and stops. And, you can define what each of these parts is so the AI has the context for when to use them.
The next, separate AI operation takes that outline and writes the full draft. I feed it the same, labeled context as I did the outline, but this time, I included the new outline in the prompt as well.
One important note here is to explicitly instruct the AI to avoid salesy language. Since it is writing as an “employee” of the company, the AI tends to be overly promotional. I give specific instructions to avoid salesy language and focus only on value and education.
The Subject Matter Expert Editor is the secret weapon to this AI content generation process.
I pass the draft to the next AI operation, which I’ve given the persona of a "Senior Content Editor" who is also an SME on the blog topic.
This step significantly improves the quality because the AI tool first looks for any gaps in the information, provides analogies to help with understanding, and catches repetitive phrasing and structural issues.
One useful trick here, too, is to use different AI tools for the drafting and SME editor steps, for example, Google Gemini to write the draft and OpenAI as the SME editor. You get the benefit of both LLM knowledge bases to deepen and enhance the value of your content.
Once the content is written and edited, the automation performs two final tasks:
Finally, the system circles back to the Editorial Calendar, pastes the link to the draft doc on my Google Drive, and updates the status to "Draft Ready."
What we have built here is a system that takes a simple topic idea and turns it into a well-researched, SEO-optimized, expertly structured first draft—all in about 3 minutes.
However, there is one fatal flaw in this specific experiment: I am not actually a Charity Golf expert.
While the AI can simulate expertise, it cannot replace lived experience. In the next post, we are going to solve this by automating the retrieval of real human expert insights to feed into this machine.
Want to build content automation yourself? I have made the entire Make.com blueprint and the Editorial Calendar template available. You can download it and upload to your own Make account using the links below.
Follow along as we transform my side hustle into a fully Ai-native business. Hopefully we all learn a few lessons along the way and I'll be sharing the plans and automations I'm building.
L2 Digital © All Rights Reserved 2025
