Pull your phone out of your pocket and open up a new text message. Type “I think we should all” and then select the option your phone suggests. And now accept the next suggestion. Continue until you have something that resembles a complete sentence.
This is called the predictive text game, and the outcomes are sometimes silly, often gibberish, and occasionally profound. You may have seen these prompts posted on social media. Sharing the outrageous results with friends can be a fun diversion. But nobody mistakes it for serious craft.
And yet, it’s not that different from the functionality of today’s AI tools.
Generative AI from Google, OpenAI, and others relies on large language models (LLMs) trained on a massive subset of the internet. This incredible resource for context provides much better predictions than texting on your Apple or Android phone. However, the central method is very similar: find the most likely next word based on the information available.
In technical writing, you may be tempted to think the right prompts will return the compelling marketing content you’re looking for. But what you’ll likely receive is an average of everyone else’s existing content. You’ll experience the Curse of the Generic, a plague that existed in technical content long before AI. Your content needs to say something different to attract the right technical audience.
Most technical content needs a point of view. Too often, this is missing—a lack of differentiation is the number one mistake made with technical content marketing.
Whether you use AI or not, you must understand why your product matters to developers and other technical audiences you want to reach. You need to frame your content around the problems your product solves. While AI has its uses, which we’ll get into, it doesn’t yet have the capability of reflecting your unique point of view.
Where Automation Excels With Technical Content
To start, we must acknowledge a potential awkward bias: this article is by a technical content agency. In other posts, we’ve stitched together countless ideas about what makes great documentation, blog posts, and marketing so seamless that developers won’t notice the hemlines. You might expect us to say AI belongs nowhere. However, there are many types of technical content. Not all of them require the same amount of opinion. In fact, some should be “just the facts.”
Take documentation, for example. Developers value accurate, updated, and clear information in docs pages. Even within these technical resources, some pages are more functional, while others are contextual:
- Functional: What’s supported and how it works (reference content)
- Contextual: How it’s used and why it matters (guide content)
There’s already a lot of automation in use for functional documentation. For example, you might generate an API reference from an OpenAPI description. Writing needs to happen here, but the point of view is less important.
Technical writer Tom Johnson shared how AI helped him accelerate his work. He distilled 100+ pages of new product information. He used AI-generated results to create outlines based on his existing documentation templates. He even used AI to help outline and create the first draft of a user guide.
Though most of the work fits into functional content, this last one is contextual. However, it still follows a familiar format—and remember, Johnson is also bringing his documentation experience to all of this effort. He’s supplying a “human overlay” to every AI request, as well as what gets returned.
Perhaps the most impressive automation example is something Johnson describes as “somewhat superhuman.” He used AI to organize hundreds of items stored across a Google Sheet and diagram tree into a single list. Finding structure within unstructured or semi-structured data is an area where machines thrive. Johnson admits his brain would have been “fried” had he tried to take this on himself.
Automation can be a fantastic use of AI for technical content. If there’s a highly templatized process, experiment with how it can help you. These opportunities are far more common in documentation than other types of technical content. Even Johnson concludes that “creative content” is not an area where he’s seeing the same AI impact on his work. And creativity is what you need if you want to reach more developers with technical content.
AI Interactions Are “Generic In, Generic Out”
Once developers have discovered your product, they expect you to explain it. That’s the point of documentation. The sort of marketing content that will attract those developers in the first place is an entirely different beast. You want developers to discover your product through middle-of-the-funnel technical topics that dig into their problems. That means you need a way to stand out from competitors who want to reach the same audience.
We’ve already seen that AI trained on what’s already out there won’t bring unique insights to your technical content. That’s up to you, your product’s positioning, and how well you can incorporate the point of view into your technical content.
Let’s check out an example from APIs (an area we know well). Perhaps you’ve identified GraphQL content that performs well for you.
Your SEO team does some keyword research.
That top keyword looks particularly juicy—it has thousands of monthly searches, big traffic potential, and should be rankable. You greenlight a comprehensive comparison of these two API technologies. Visions of traffic growth dance in your head as you assign this to a freelance writer or attempt a content brief with AI.
As a test, I asked AI to build this outline. It came back with something that’s not bad. Except, it’s exactly what you’d get if you asked AI for this outline. And it’s exactly what your competitors will get when they pursue the same topic.
Go ahead and try for yourself, Prompt Engineers:
Please create an outline for a post that will rank for these keywords: graphql vs rest
It’s missing any hint of perspective because you’ve given it none.
Again, let me emphasize: This is not just an AI problem; it’s a technical content problem.
The Curse of the Generic befalls all who do not take a stand and bring their viewpoint to blog posts, guides, and tutorials. In Technical Content Strategy Decoded, I suggest your job is to translate developer problems to your developer product. To set yourself apart, you must include opinions influenced by your positioning. In other words, apply a human overlay even during the planning process to create something unique.
In the case of our API architecture example, you might:
- Emphasize the flexibility of GraphQL
- Warn developers not to expose their data model
- Stress the importance of API design regardless of architecture
Any of these, or another angle, would impact the outline and eventual content. Even better, it would differentiate your post from all the other approaches to this comparison. You’d avoid the Curse of the Generic and have a much better chance to rank for the search term.
Finally, when developers read a post with your unique insights, they’re much more likely to stick around and sign up for your product.
Avoid Generic Content, with or without AI
The potential of AI is impossible to ignore, even for those with experience creating great, standout content. EveryDeveloper uses it strategically, mostly as part of initial planning. Whenever we use AI, our ideas remain the driver in the process. Identifying what’s unique or important is still primarily a human activity.
By all means, experiment. Whatever you take from AI must be questioned and shaped further. Anything you give it must include original insights if you want anything worthwhile to come out the other side. When it comes to producing impactful technical content—the sort that will attract and convert developers and others—your point of view should be at the forefront.
When you need another set of brains to help identify what’s unique and important—what matters—you can reach out to the humans of EveryDeveloper.