How We Run a Content Business With No Writers: AI Automation From Idea to Published

Most content teams have the same problem: too many channels, not enough time, and a backlog of half-finished drafts rotting in Notion. The average blog post takes 3-4 hours to produce. Multiply that by the volume you actually need to compete on search, and you're either hiring a small army or falling behind. We chose a third option.

We run a content business with no writers. Not because we cut corners on quality — but because we rebuilt the entire pipeline using AI automation. From topic selection to a live published post, the whole process runs with minimal human intervention. Here's exactly how it works.

Step 1: Topic Selection Without Guessing

The pipeline starts with a trigger — either a scheduled weekly run or a manual input when we want to move fast on something timely. We pull keyword data from a connected SEO tool (we use a lightweight API integration with Ahrefs), filtering for terms with decent search volume, low-to-medium difficulty, and clear commercial or informational intent that matches our content categories.

That filtered list feeds into an n8n workflow. The automation scores each topic against a set of rules we defined: does it fit our niche, has it been covered recently on the site, and does it align with a product or service we want to drive traffic toward? The top-ranked topic gets selected automatically and passed downstream. No editorial meetings. No spreadsheet debates. The system just picks and moves.

Step 2: AI Drafting With Actual Structure

Once a topic is selected, the workflow sends a structured prompt to the OpenAI API. The prompt isn't just 'write a blog post about X.' It includes the target keyword, the intended audience, the tone profile we've defined for the brand, a required word count range, and a content structure template — intro, H2 sections, practical takeaways, closing CTA.

The output comes back as a structured JSON object, which makes the next steps cleaner. Each section is a discrete field we can route individually. We run a second pass with a refinement prompt that checks for filler phrases, over-used AI language patterns, and anything that sounds like a press release. The goal is copy that reads like it was written by someone who actually knows the subject — not a language model trying to sound smart.

We're not precious about AI involvement. The draft is AI-generated. The system, the prompts, the rules — that's the real IP.

Step 3: SEO Optimization and Publishing to Static Site

After drafting, the workflow automatically generates a meta description, pulls in the target keyword for on-page placement checks, and formats the post for our static site builder. We use Astro with content stored as Markdown files in a GitHub repo. The automation commits the new file directly via the GitHub API, which triggers a Netlify build and deploys the post live — no manual publishing step required.

Internal linking is handled by a lookup step that checks existing posts for relevant anchor opportunities and injects two or three links before the file is committed. It's not perfect, but it's consistent, and consistency at scale beats occasional perfection. The post is live, indexed-ready, and formatted correctly without a human touching a CMS.

Step 4: Distribution and Performance Logging

Publishing is not the end of the workflow — it's the middle. Once the post is live, the automation pulls the URL and generates a short-form version of the post for X (formerly Twitter). This gets posted via the X API with relevant hashtags appended automatically based on the content category. LinkedIn follows a similar pattern with a slightly longer format prompt.

The final step logs the post details — title, URL, target keyword, publish date, and distribution status — into an Airtable base. We have a separate scheduled workflow that checks Google Search Console data weekly and updates those rows with impression and click data. Over time, this creates a performance feedback loop. Topics and formats that perform well get weighted higher in future topic selection. The system learns what works and skews toward it.

The full stack runs on n8n, OpenAI, GitHub, Netlify, Airtable, and the GSC API. Nothing exotic. Total workflow build time was around 12 hours. It now produces and publishes multiple posts per week without manual input.

If you're still writing every post by hand, you're not competing — you're just staying busy. The businesses winning on content right now aren't the ones with the best writers. They're the ones with the best systems. The tools exist. The APIs are cheap. The only thing missing is someone willing to build the workflow instead of opening a Google Doc.

We help companies build exactly this kind of infrastructure — custom AI content pipelines designed around your stack, your brand, and your growth targets. Not templates. Not off-the-shelf tools. Actual systems that run.