My AI Content Workflow: The Tools I Use to Write and Publish Every Post

❤️ Before we get started I'd like to thank you for using my affiliate links to sign up to free trials, LLMs are constantly stealing my content and you help me stay afloat and create more of this content to AI enthusiasts and small business owners. For example if you use my Airtable link I get Airtable credits and can create even more content for you ❤️

And before someone who tests AI tools every day tells you her secrets, let me give you a bit of context on where I'm at right now with my AI workflow and overall. We relaunched the website on Webflow in mid-December, after building out the infrastructure, data, and design since around October. When we were finally ready to move from Squarespace (not recommended) to Webflow (highly recommended), we had to transfer all the information about AI tools, page content, and more into our current AI-powered brain: Airtable.

So I've been working in this new environment for about two months now and it's still a work in progress. I'm not going to pretend everything runs seamless and that I just push a button to generate perfect output. But we're getting closer to what I think is the best possible combination: unique human insights, enriched and improved by generative AI.

That said, it's already a 100% improvement on what I was doing before and I'm genuinely excited about what we'll be able to build next, because the best part is not having to do all those tedious little jobs anymore. In this post I'll walk you through the tools I use, the structure of my AI content workflow, and what I'm planning to build next.

The Tools I Use for My AI Content Workflow

Airtable is my database and the backbone of my entire AI content workflow. Think of it like an Excel sheet on steroids where tables can be interconnected and dynamic. It stores basically all my data and is where a lot of the automation happens, because Airtable connects to various AI models like GPT and Gemini. You can also connect it to other providers like Straico, which bundles multiple AI models in one place.

Webflow is my AI-driven website builder and it allows endless scaling thanks to its CMS collection and template setup. Via a script and various automations, we send the data from Airtable straight into the Webflow CMS. Every page, tool, and media file has its own table in Airtable and its own CMS collection in Webflow. It is super customisable and adding rules and filters makes it even more capable for updating and managing content efficiently.

NeuronWriter is my SEO tool. It analyses competitor websites based on a keyword and gives me around 20 basic keywords and 40 extended ones. It also tracks my optimisation score based on how well I use those keywords across the main copy, headlines, meta titles, and descriptions.

Claude is my reliable co-author and favourite AI model. It used to be ChatGPT, but Claude's ability to ask questions back and really understand context makes it a much better fit for my content creation process. (And no, that last bit was actually written by me, not smuggled in by Claude, hehe.)

My AI Content Workflow Step by Step

I've already gone into the automations in more depth in my post "How I Automate my Website: Airtable Webflow Integration", but here I want to dive deeper into the workflow itself and how I use AI at each step of the content creation process.

Idea and Keyword Research

I usually start with an idea and brainstorm titles and keywords together with Claude. It's not able to research keyword volume data yet, so I do that part in NeuronWriter and feed the results back into Claude. In this combo, I'm definitely the one in the lead. Claude is enthusiastic about topics with very low search potential, whereas I as a small business owner want to be tailoring content to find a real audience. Where Claude leads is in suggesting the perfect title and automatically giving me the character count, because I've built a skill that defines my workflow and preferences upfront.

Writing the Blog

Believe it or not, I still write all my blogs myself. I think about the content and form my own unique insights, because I genuinely don't think content will rank if it isn't unique. There are so many websites competing with other mediocre websites full of AI slop and I don't want to be one of them. So I write the first draft myself without worrying too much about nice formulations or keyword placement, because at that point I hand over to Claude.

I think this is the perfect allocation of labour: my brain does the critical, unique thinking and Claude does the beautiful formulating. It does an incredible job of keeping my personal tone of voice and even leaving in my silly jokes. At times I let Claude summarize everything in the end for a conclusion, but that's as far as it gets. Once that's done, the polished content goes straight into Airtable.

Letting Airtable Handle the Blog Header

Depending on the topic or format, a lot of small automations kick in within Airtable. A field agent, which is basically a mini automation within a single field, takes care of creating the header image automatically based on the blog title. Airtable can also integrate with AI image models and handle this task entirely, I just have to fine tune the prompt well.

For comparison pages, Airtable will research all the key features, prices, and free trial details for me automatically. There is still a lot of room for more automations here, but I want to get used to this workflow first before I overengineer everything. I can always overengineer it later. Or let my team do it. Either way, someone will.

Adding the Data to Airtable

For now I'm still doing this part manually. I haven't fully let go of the leash with Claude yet because I've heard too many stories about AI agents going rogue. But I feel like our relationship is progressing and we're getting close to a more serious stage where I can introduce Claude to the parents. In this case: Airtable.

Publishing to Webflow

As mentioned earlier, a script sends the data over to Webflow via an automated process. These scripts can be quite elaborate, sometimes running to 1000 lines depending on the page type. I'll be honest, I didn't set them up myself. My team built them because there is just too much that can go wrong. I can fix smaller issues together with Claude, but I've had enough headaches to know not to mess with them beyond that.

Why This Works for a Small Business

What used to require a virtual assistant, a real human, is now handled by me, Airtable, and Claude. It has made me so much faster. Instead of explaining everything, writing tickets in ClickUp, and waiting, I can just implement things myself and get content online much quicker. In a field like AI tools where something new drops every single day, that speed is essential.

And this hasn't meant getting rid of my assistant. It means we can finally tackle areas we never had the bandwidth for before. For us that's social media. We now create content for Instagram, Facebook, and more, which simply wasn't happening before. We even have an automation set up for that: whenever I create a new page or add a new tool, an automation generates a summary and writes a first post draft straight away.

My assistant can now focus on video editing, which is honestly much more fun anyway.
I don't believe in downsizing as a small business. I believe in staying lean but getting more efficient, and this AI content workflow has made that possible.

Next Steps for my AI solutions

Once I've gotten to know the ins and outs of Airtable better can customize automations better, the goal is a proper domino flow where one step triggers the next, with human checks along the way. We're not quite there yet. We're still making changes to the front end of the website, so it doesn't make sense to build a tight automation sequence when I'm still moving the first domino or change the input constantly. But we are building more and more tools to streamline the process already today, like an image downsizer in Airtable that automatically compresses images to under 100kb.

One of the things I'm most excited about is building my own keyword cannibalisation app with Claude. It will pull data directly from Google Search Console and flag where cannibalisation is happening. No more manual data analytics, which I absolutely hate.

And I think that's exactly how clever small business owners should use AI to create: let it do the things you hate doing. I hate data analysis, so that's Claude's job now. They don't mind. I do love generating my own insights, so I'll keep doing that part myself. I'm the marketer and blogger, Claude does the rest. 
Soon I'll also connect Claude directly to Airtable so blog content can be deployed there straight from Claude, cutting out the manual step entirely.
After that, the plan is to give Claude access to my emails so it can handle backlink requests directly, and maybe even implement them itself. Honestly, another thing I'd rather not be doing.