Content Marketing in the Age of AI: How to Write Better Content, Faster

There’s a reason that everywhere you turn, you’ll bump into something that’s “powered by AI”. That’s because, no matter how you feel about artificial intelligence, there’s no denying that it’s one of the most revolutionary technologies in human history.

To some, it might seem like the learning curve for leveraging AI is steep. Fortunately, for most use cases, it isn’t - thanks in large part to the latest advances in large language models (LLMs) like ChatGPT, Grok, Claude, and more.

For marketers and business owners in particular, one of the easiest (and most fruitful) use cases for these LLMs is content marketing. With the right approach, you can spin up content that isn’t just fast to create, but also sounds like your brand, and most importantly, adds actual value to your target audience.

In this post, I’m going to walk you through the key strategies I’ve developed over the years in working with LLMs for content marketing. By the end, you’ll have a better understanding of not just the specific prompts I use to get good results, but a new way of thinking that will help you write better content in less time.

The Foundation: Content Marketing Fundamentals

Since this guide is about content marketing, let’s first get on the same page about what it is, what it’s for, and why you need it.

In my experience, content is one of the most effective ways to build credibility as a brand.

Obviously it’s important to have a great offer, a well-defined target audience, and a strong unique selling proposition. But even with those pieces in place, many potential buyers won’t be convinced without more supporting evidence.

Moreover, content marketing is the best vehicle for providing results in advance by sharing your expertise for free.

The dirty little secret behind why some of the biggest names out there give away their best stuff for free is that it doesn’t just showcase their expertise, it also triggers reciprocity in the consumer.

When you arm people with actionable insights that can actually help them solve their problems, many will have a tendency to feel grateful for your generosity. That one-two punch of credibility and generosity helps increase the likelihood that they’ll do business with you, and not your competition.

Content can come in many forms, from quick video tutorials as YouTube shorts to technical white papers, and everything in between.

As long as you’re sharing something that your ideal buyer will find valuable, you’re winning at content marketing.

And the best news of all? ChatGPT can help make this process more efficient - once you know how to overcome its default settings.

The War Against “GPT Default”

Before we dive into prompt engineering specifics, we’re going to take one more quick detour to understand some of the limitations of LLMs, and why they end up resulting in substandard writing.

This will help explain some of the logic that I’ll go deeper into over the course of this guide, and (ideally) help you feel less frustrated when ChatGPT spits out a pile of hot garbage.

Author’s aside: From here on, I’m going to make reference to ChatGPT as a proxy for the range of LLMs that are out there on the market. They each have their strengths and limitations, but ChatGPT is by far the most popular - not to mention it’s the one I use most often.

The way I like to think of ChatGPT is as an enormous pattern recognition machine. As you may know, it’s been trained on an astronomical amount of data, with the goal of being able to use these data to help users with everything from coding to image generation - and of course, writing.

The fact that ChatGPT has processed all of this information is simultaneously its superpower and its greatest weakness when it comes to content marketing.

Its strength lies in the fact that it gives you on-demand access to deep research and insights, even about complex topics.

The weakness is that, by its very nature, it tends to generate responses that are based on a law of averages.

For example, if a certain sentence structure appears over and over in its training data, it’s going to assume that this is the best way to write that sentence - the result being a style that I like to call “GPT Default”.

GPT Default is the type of writing that doesn’t just get flagged by AI detectors (which is bad for SEO), it’s also obvious to most readers. If you’ve ever seen a string of bland, uninspired LinkedIn comments, you know exactly what I’m talking about.

This isn’t just a problem at the level of the sentence structure, either.

GPT Default can become a stumbling block at every level of the content creation process, from ideation through the final draft. Fortunately, there are some ways to get around this default response, and turn it into something truly valuable.

That’s our goal for the rest of this guide, starting with the ideation process.

How to Prime ChatGPT for Content Ideation

If you’re going to create content that your readers will love, the first thing you’ll need is some ideas. Ideally, you want to create content that lies at the intersection of “what they want” and “how your brand can help”.

That’s why, if you want ChatGPT to give you better content ideas, you first need to feed it as much information as you can about these two sides of the equation.

Fortunately, ChatGPT can process this content from a wide range of formats. If you already have a website, all you have to do is copy and paste the relevant URLs, and it can scan those pages and commit them to memory. This is a great way to have it develop a better understanding of your brand, your offer, and to some degree, your target audience.

I’ve fed ChatGPT information from .docx files, text files, PDFs, and even screenshots - and it’s done an excellent job of reading, understanding, and drawing insights from all of this source material.

The same goes for having it understand your target audience. If you’ve put together any persona development documents, for example, you can upload those to chat, and have it add those insights to its working memory.

As with other steps in the content creation process, one of the keys to success is to quiz the AI at strategic intervals to see how well it understands what you’ve shared.

Ask it to summarize, share the most important points, or even call out areas where you might be lacking clarity. You can then refine its responses based on your direct experience, so that everything downstream stays on brand and on target.

Once you feel like it has a good grasp on your brand, your audience, and how you help them, it’s time to have it start cranking out content ideas. The more specific you are with your prompt, the better.

So if you want in-depth blog post ideas that target specific keywords, don’t just ask it to “give you content ideas”. Tell it the list of keywords (which again, you could upload from a spreadsheet or doc) and how long you want the posts to be. Conversely, if you want ideas for fun social posts or quick video tips, say so.

As you give it feedback on the quality of those ideas (or lack thereof), it will learn about your preferences, and start to get better at generating responses that align with your expectations.

Once you have your list of ideas in hand, I strongly recommend porting it over to a working document for future reference. Despite its computing superpower, ChatGPT does have a tendency to get amnesia, which then forces you to have to scroll back up through the chat to find what you’re looking for - not fun, believe me.

Anyway, let’s assume that at this point you’re happy with your list of content ideas. From here, it’s time to pick your favorite one and start the next phase: outlining.

Outlining With AI: Tips and Tricks

When you use ChatGPT to create a content outline, the process is usually straightforward. Assuming you were specific enough in what you want to cover, and have given the AI enough background information, it usually does an admirable job at organizing information in a logical way.

That said, those same dangers of GPT Default can crop up here as well, and may be harder to spot.

The first issue is redundancy. As the AI creates the outline, keep a close eye on the topics (typically in bullet point format) that it includes in each section of the piece.

In my experience, ChatGPT prioritizes giving every section equal weight, in terms of how much information it should include. The result is that it will sometimes repeat what’s already been discussed in a previous section, but with a slightly different angle.

Another issue that you will sometimes run into is a lack of logical flow to the piece. This is less common, but still something to watch out for.

In both of these cases, you can go one of two routes to fix the issue:

The first is to just use your own common sense and tell it the specific issues you found in the outline.

The second is to ask it to objectively look through what it wrote, and see if it can identify any issues. Specifically, I ask it to find redundancy, lack of logical flow, topics that don’t belong, or critical information that’s missing.

Both of these methods typically yield good results, but again, you have to be the final judge as to what makes sense for the reader.

Last but not least, you can always ask ChatGPT to give you more or less detail in your outline. For more in-depth pieces, I sometimes have it write a brief introductory sentence to summarize the purpose of the section, then follow that with bullet points on the key topics to cover. For shorter pieces, it may be fine to just have the name of the section with 2-3 bullet points.

As before, I recommend copying your approved outline over to a separate document for easy reference. Now that you have it in hand, it’s time to get down to the business of writing.

How to Write a Killer First Draft With ChatGPT

There are a few different approaches you can take when converting your outline to a fully written piece. Which one you choose depends on a combination of personal preference and the length of the content.

If you’re writing something that’s highly technical, requires a lot of citations, or needs to be longer than 2,000 words, one powerful option is to use the Deep Research function. This is available with paid subscriptions, and is well worth the investment if you’re regularly writing in-depth content pieces.

The only caveat with this approach is that, in my experience, it tends to create content with a bloated word count. This ends up shortening the time spent during the writing phase, but makes the editing phase longer than usual. That said, if you have the Deep Research tokens available, this is a good way to spend a few of them to see how well they perform.

If your content is on the shorter end of the spectrum - say, under 1200 words - then it’s worth trying to have ChatGPT write the whole thing in one shot. This has the obvious advantage of a faster writing process, since you can kick out the entire first draft in a matter of seconds.

For most content, I use a third approach, which is to work through an initial draft of the piece, one section at a time. The main advantage of this approach is that, once it gets past the introduction, you can make real-time adjustments to the tone and formatting, so that the sections that follow are closer to the mark.

To keep the AI from losing track of where we are in the piece, I will sometimes even copy over the outline section itself, to remind it of what we’re supposed to cover. This can make the editing process faster and easier, which is, after all, the whole point of this exercise.

If you’d like, you can also give ChatGPT real-time feedback on how it’s writing the initial draft. While I don’t recommend getting too far into the weeds at this point, if it’s missing the mark badly from the get go, you might as well have it make adjustments up front.

Last but not least, you can specify to ChatGPT whether you want the piece in question to be written in chat or written in Canvas, which was introduced in October of 2024. Personally, I find that the editing process is easier when I keep things in chat, and port them over to my working doc one section at a time. That said, Canvas has its advantages as well, so I suggest experimenting with different workflows until you find what works best for you.

Make it Shine: How to Get a Great Final Draft

Just like when you write content without AI assistance, it’s completely fine to have a not-so-great first draft. The interesting thing about the editing process for AI-generated content is that, unlike the wide variety of writing quirks that humans might have, its issues are relatively predictable.

Recognizing and fixing these issues is a skill that gets better through practice and repetition, but the good news is that if you have any intuition as to what constitutes “good writing”, then you can direct the AI to fix its own mistakes. This will typically get you anywhere from 85-100% of the way to your final, polished content piece.

In terms of how to think about editing AI-generated content, it’s the same as the rest of the outlining and writing process: we’re moving from broad strokes down to ever-finer details. To help illustrate this point, let’s look at a few specific problems to watch out for, and some prompt ideas to help correct them.

Common AI Writing Issues to Watch Out For

The first thing to double check is that ChatGPT didn’t lose track of any of the critical information that was agreed upon during the outlining process. Again, temporary amnesia seems to be all-too-common, which is why we copied the final outline over for easy reference.

Along these same lines, it’s always good to double check for redundancy across the entire piece, to make sure you’re not going over the same information twice - unless you’re doing so for emphasis or stylistic reasons, for example.

From here, you should be on the lookout for patterns in the writing that hint at AI-generated content. ChatGPT likes to land on a given formula that it thinks will satisfy the end user, then stick to it like glue across the entirety of the piece. Not great for keeping readers engaged, and even worse for triggering AI detectors.

One way you might see the AI being formulaic is in the length and structure of the content section itself. Let’s imagine that the first main section of the piece consists of (1) an introductory paragraph of three sentences, (2) a series of 3 bullet points at a specific level of detail, and (3) a conclusion that transitions into the next section. Unless directed otherwise, you can bet your bottom dollar that every section thereafter will follow this exact formula.

This tendency is just as obvious in how those paragraphs are phrased. To drive this point home, I had ChatGPT write a quick article about different aspects of getting in shape. Here are the introductory sentences for some of the different sections of the piece (with emphasis added):

“Getting in shape doesn’t require fancy gear or the perfect workout playlist—just consistency and a basic understanding of what your body needs.”

“Flexibility doesn’t get the same spotlight as cardio or strength, but it’s just as important.”

“Strength training isn’t just for bodybuilders. It’s key to maintaining muscle mass, supporting joint health, and increasing your resting metabolism—meaning you burn more calories, even when you’re not moving.”

“Getting in shape isn’t about chasing perfection—it’s about building a foundation you can stick with.”

As you can see, every single one of these intro sentences follows a formula of “negative framing” followed by some more positive or factual statement. This is one example of formulaic writing, which is one of the two major stylistic issues you’ll face. The other is tone, which is more subtle - and is also where it starts to make sense to move on to a manual editing process.

Final Edits: Automatic or Manual?

The cool thing about most of these mistakes that we’ve identified so far is that, again, the AI can typically recognize and fix them. One way to get it to do this is by uploading a PDF or .docx file with the current version of the draft, then asking it to objectively analyze it. Try entering prompts such as:

“Are there any areas in this piece where information is redundant?”

“There is some formulaic writing between the second and third sections of the piece. How might we fix those so that they don’t follow the exact same sentence structure?”

“This has too many em dashes, most of which are not needed. Please remove them wherever possible and rephrase the sentences accordingly.”

“This tone is too formal in places. Please make it more conversational, without becoming too casual.”

This is just a sample of the kind of feedback that will make your first draft better, without you having to make any manual edits. That said, even with the most sophisticated prompt engineering has its limitations. ChatGPT still doesn’t have much in terms of style, for one thing.

Another issue to keep any eye on is how much time you’re investing in using prompts to make your edits. At a certain point, you know what you want to say, or how to say a certain phrase in a more natural, human way. Oftentimes those final tweaks are both better and faster when done by hand, so don’t be disappointed when ChatGPT can’t figure out how to get it perfect by prompt engineering alone.

Conclusion: For AI-Generated Content, Process is the Key

ChatGPT is like any other tool (software or otherwise) in that the more you work with it, the more proficient you get. The LLMs themselves have also been improving at a stunning rate, and they’re not likely to slow down anytime soon.

These two factors alone should be enough to convince you that if you haven’t been leveraging AI for your content marketing, now’s a good time to start. It’s part of my workflow almost every day of the week, and there’s zero doubt that it has more or less doubled my creative output.

That’s partly due to the faster creation process, but just as important is the fact that it reduces the amount of thinking I have to do to make something great.

If your marketing campaigns are in need of more content to build brand authority, support email campaigns, or generate leads, and you’re still not sure where to start, feel free to request a free 30-minute consultation. I’m always happy to share new ideas and help give direction on how to get better results from your marketing.