A Guide to Building Better Blog Posts with AI

Can AI help us craft better blog posts without falling into the trap of producing generic “content”? Absolutely. In this post, I’ll share how I leverage large language models to enhance the quality and efficiency of the blog posts you find here. Build Better Blog Posts with AI!

There’s a common misconception that you can simply throw a prompt at an LLM like ChatGPT and instantly generate high-quality content. That’s far from the truth.

While LLMs have the potential to create reasonable text, they can just as easily churn out bland, repetitive, and uninspiring material if not used thoughtfully. In my approach, I treat LLMs as highly capable assistants, handling some of the groundwork while I focus on shaping and refining the final product.

Here’s how I do it:

Start with the Outline

Building Better Blog Posts with AI requires some thought to be devoted to your workflow. To demonstrate my process, I’ll walk you through how I wrote a blog post titled “In-Source versus Out-of-Source Builds.” The first step is to create a clear outline. This outline will guide the structure and flow of the post, helping me stay focused on the key points I want to address. Below is the initial draft of my outline, which serves as the foundation for the article and the approach I’ll take to explore the topic.

# In-Source vs. Out-of-Source Builds

## Brief Introduction
- Why I care, and you should too

## Definition of In-Source vs. Out-of-Source Builds
### In Source Builds
- What they are
- What systems default to these

### Out-of-Source Builds
- What they are
- What systems default to these

## My Search for Help
- Looking for help in 2018 on Stack Overflow.
- Why I needed help.
- Since viewed over 10k times.

## Authoritative Advice
- Why I can give authoritative advice here
- Avoid in-source builds

## Conclusion

With my initial outline in hand, I asked the LLM for suggestions on how to enhance it. This step is crucial because it allows me to tap into the LLM’s vast knowledge base. It allows me to uncover ideas or angles I might have missed.

By doing this, I’m not just looking for surface-level improvements. I’m seeking to refine the structure, identify any gaps, and explore additional perspectives that could add depth to the article. The LLM can provide fresh insights. It can also suggest alternative ways to organize the content, or even highlight key points that could be expanded or clarified.

Refining the LLM’s Outline

While I don’t rely solely on the LLM’s suggestions, I find that this process often brings up insights that can enhance the overall quality of the post. It’s a collaborative effort. The LLM acts as a brainstorming partner, helping me think more critically about the content and structure. This interaction ensures that the final product is not only thorough but also engaging and well-rounded.

That said, the suggestions provided by the LLM aren’t always perfect. Generally, the GPT offers great ideas, but it’s not uncommon to make a few tweaks to its outline. Often, this involves trimming any unnecessary “fluff” that might dilute the message. It also involves making adjustments to maintain the focus and clarity of the article. The LLM can offer a broad perspective, but it’s up to me to refine its input. This ensures the content remains sharp, relevant, and aligned with my original vision. This balance between AI-generated suggestions and human judgment is what ultimately leads to a stronger, more compelling blog post.

Writing the Article

Building Better Blog Posts with AI requires more than a great outline. Now that the outline is complete, it’s time to bring the article to life by filling in the details. This is where the human element truly comes into play.

While the outline provides a roadmap, the actual writing process involves adding depth, insight, and personality to the framework. I start by expanding on each point in the outline, using my knowledge, experience, and unique perspective to flesh out the ideas. This is the stage where the article transitions from a skeleton of concepts to an engaging article.

During this process, I rely on my own voice to ensure the writing feels authentic and resonates with the reader. It’s not just about presenting information. It’s about telling a story, making connections, and offering valuable insights that go beyond the basics. I carefully consider the flow of the article, the tone I want to set, and how each section will engage the reader and keep them interested from start to finish.

While I may use AI tools to assist with certain aspects, like refining language or generating ideas, the core of the writing remains a human endeavor. This is where creativity and critical thinking are essential. They allow me to produce content that is not only informative but also compelling and original.

Authoring Code Examples

Especially in a technical blog, building better blog posts with AI can be used to generate compelling code examples. This is one area where large language models (LLMs) can truly shine. They can quickly produce snippets of code that illustrate key concepts or demonstrate best practices. However, while LLMs are great at generating code, they can’t build it to know that it actually compiles or even works! That’s where the human touch becomes essential.

Testing Code Examples

After the LLM generates a code example, my next step is to thoroughly test it. An LLM might produce code that looks correct, but without testing, there’s no guarantee that it actually works. I use tools like Compiler Explorer to compile and run the code, ensuring it functions as intended. This step is crucial not only for accuracy but also for maintaining the credibility of the blog post.

During testing, I also refine the code. This might involve updating comments for clarity, optimizing the code for better performance, or making it more readable for the target audience. I also consider whether the example could be improved to illustrate better the concept I’m discussing. The goal is to create code that is functional, easy to understand, and relevant to the readers.

Finally, I generate a link to the working code example in Compiler Explorer. I usually include this link in the blog post as well. This gives readers a quick way to see and experiment with the code themselves, adding an interactive element to the post.

Copyediting with AI

I have a genuine love for grammar. (A fact that would probably surprise my 12-year-old self, who dreaded diagramming sentences (and to this day does not understand why I had to do that.) Over the years, I’ve studied guides like Strunk and White and Writer’s Inc, and I can easily switch into “editor” mode when needed. However, with the advent of modern tools, I don’t need to rely on my own skills.

Enter Grammarly—a tool that has become an invaluable part of my writing process. While I enjoy the nuances of grammar, I use Grammarly to ensure that everything I write is polished and precise. It helps catch any mistakes I might overlook and offers suggestions to refine the text, making it more straightforward and engaging.

Grammarly serves as a second set of eyes, helping me maintain a high standard of writing without getting bogged down in the minutiae. It’s handy for catching those subtle errors or awkward phrasings that can easily slip through when you’re deep in the writing process. Plus, it provides suggestions for improving the overall readability and flow of the text, which is crucial for keeping readers engaged.

Here’s an example of how Grammarly reviews text for the blog, highlighting areas for improvement and offering actionable suggestions:

Building Better Header Images with AI

When I first started my blog, I wanted all of my header images to follow a cohesive theme. Initially, I opted for nature photos loosely related to the blog’s topics. However, I soon found this approach to be uninspiring and lacking the unique identity I was aiming for. That’s when I decided to turn to AI once again, this time to create consistent and engaging header images.

I developed a custom GPT specifically designed to generate these images. The concept was to feature a recurring character—a female programmer, loosely based on my daughter—who would serve as the visual centerpiece for my blog. In each header image, this character is depicted working in a futuristic environment, solving complex engineering problems. This adds a personal touch to the blog and gives it a distinctive and cohesive aesthetic.

With this setup, I use the GPT to create the banner image for each blog post. The AI helps me maintain consistency while also allowing me to explore creative and visually appealing designs that align with the blog’s technical themes. The result is a series of unique header images that tie the blog together and enhance the reader’s experience.

Though the GPT is great, I usually have to generate several variants of an image, trying different prompts to refine the image. Here are some of the images that were generated for the blog post:

Final Thoughts on Building Better Blog Posts with AI

“So you use AI to write your blog!” – Yep!

And it makes it a much better blog than it would otherwise be. We really can be Building Better Blog Posts with AI! AI is a tool that can help refine ideas and accelerate their delivery. It isn’t evil, nor does it diminish the value of the content. To the contrary, properly leveraged, AI is can add significant value to the work we produce.