Eight unique ways to use AI for content

Introduction

Imagine you’re a weaver in 19^th-century England, producing a few bolts of cloth a year for the local landowner. It’s a skilled job and you earn a living wage; your mother and grandmother were all weavers, and you expect to pass on the skills to your daughter.

The next day your boss unboxes a strange machine from France. This “Jacquard Loom” will do the job of four weavers, and do it faster – which means your life is about to change. 

So you have two options. Do you become a “Luddite”, and join the mob smashing the machines? Or do you see an opportunity – to rethink, retrain, and use it to increase your own productivity?

You can guess which newer technology this riffs on: it’s AI, and how Large Language Models (LLMs) are generating content at scale. To some it’s fascinating. To others, exhilarating. And some find it just plain scary. But whatever you think, it’s fast becoming part of the US$400bn content marketing landscape – and it’s here to stay.

As a content creation platform, Contentoo is big on AI. But we don’t see it as a replacement for our writers. (Even our largest-scale documentation and localisation projects have flesh-and-blood humans firmly in the loop.) We see AI as a toolbox of techniques – a view shared by technology copywriter Chris Worth. Just as the Jacquard Loom enabled weavers to produce intricate designs that sold for higher prices, copywriters like Chris are using LLMs to develop innovative new ways of making content more effective.

Chris writes for many Contentoo clients, with a near-perfect NPS of 97. He’s also an AI pro, with certifications in Prompt Engineering and Python. (He created an easy guide to the skills levels of AI-savvy copywriters: download it here.) In this article, Chris shares some of the ways he uses AI to improve his own work – and how you can, too. 

Disclaimer: It’s not meant to actually write content from scratch, LLM’s are terrible at that, but rather to improve processes and creativity. 

No code, no problem: the plain talking of LLMs

The beauty of AI is its natural language interface: even a complete beginner can open a context window and start interacting, asking questions and delving into subjects. And the six techniques below are no different. None require any coding skills, and any writer can apply them to save time and make their content more meaningful.

Of course, Contentoo is a (mostly human powered) technology platform, so after those six you’ll see some more advanced concepts, too. But let’s start with these six tips for adding value to your text with a dash of AI.

Technique #1: use AI to explore your target’s mind

No matter how well you know your client, you’re not able to read their mind. But being able to would really help, wouldn’t it? They’ve spent years developing products and talking to customers; they’ve put their heart and soul into understanding their market, and they deal with a thousand worries every day. This gives us one of the most useful ways to use AI – asking it to behave as your target audience.

SAMPLE PROMPT: Act as a [job title] in a [company type]. You are the target audience for [product], whose value proposition is [VP, USP, or key benefits]. What are the top five issues that keep you awake at night which this product could solve? In your answer include examples of how this product would address these issues from the perspective of this prospect.

Note this technique starts you off on the right foot: forcing you to think about customer benefits, not product features. Remember: people don’t buy drills, they buy holes. 

Technique #2: come up with a concept for your content 

Every article worth reading has a “hook”: some interesting fact or finding that makes the reader smile. And LLMs are a great source of hook ideas. Writers spend hours wracking their brains to find that splashy thought that draws the reader in – so ask your friendly AI instead.

SAMPLE PROMPT: I am writing an article about [subject] for a company that makes [product or service]. Give me a surprising fact about this subject that not many people know. Feel free to use analogies or metaphors to relate this fact to readers’ everyday experiences.

Using this prompt for a roller bearing client gave this output: “The precision steel balls used in these bearings are so perfectly round that if they were scaled up to the size of the Earth, the largest variation in their surface would be no more than 2.5 metres.” Now, isn’t that a great way to start an article?

Technique #3: apply a client’s Style Guide and Tone of Voice

Every content creator with more that one client spends hours and hours making sure their work complies with the client’s language and style preferences – and AI can do this work for you in minutes. You’ll need a copy of the relevant client documents and of course your own article, either pasted into the context window as text or uploaded as an attachment. When using this prompt, always check no sensitive client data appears in either file.

SAMPLE PROMPT: File [name] is an article written for a company. File [name] is the company’s Style Guide. Analyse the Style Guide then apply it to the article by inserting the characters >>> where the article does not follow a Style Guide requirement, noting the requirement after the >>>. Output results as a downloadable file.

That “output results as a downloadable file” is important; it gives you a new file to work from, with any questionable content marked for correction. Try it!

Technique #4: create the ideal framework for your content

Most forms of writing follow recognised structures: everything from the Hero’s Journey in fiction to the AIDA sequence of direct marketing. It’s copywriting 101. Hence another AI pro tip: use it to give you a structure to write to.

SAMPLE PROMPT: I am writing an article of type [blog, ad, brochure] about [subject] with the purpose of [reason for writing it]. I want it to follow the [name of framework] structure from [source of structure, for example a marketing textbook]. Give me a copy plan for this article based on this structure.

There’s a bonus technique here: why not use a similar method to check your work when it’s done? Upload your work back into the AI’s context window, then prompt the AI to check it complies with the structure you’re aiming for. Even more fun: ask the AI if it can guess the structure you used!

Technique 5: check for inconsistencies and contradictions

We’ve all done it on a long piece: made an assertion early on that wasn’t supported by later data, or used two evidence points that said different things. And of course you’ll only notice them after submission … or worse, the client sees them first. That’s another great use for AI: upload your work and ask it to check and challenge your conclusions. 

SAMPLE PROMPT: File [name] may contain inaccuracies, unwarranted assumptions, unsupported conclusions, and contradictory evidence or other errors. I want to eliminate these problems and make the text stronger. Analyse this file in depth and point out parts I should correct, change, delete, strengthen, or explain further.

Here you’re using the AI as your Devil’s Advocate: deliberately asking it to find the holes in your argument that make it weak. But be warned: AI can be very critical when you ask it to be.

Technique #6: Source evidence to support your arguments

Much as we love AI, Large Language Models make mistakes. Remember that ultimately an LLM is not “answering” your question; it’s generating a series of word fragments (called tokens) with a statistically high likelihood of looking like an answer. 

While they’re incredibly useful, this also limits their scope. If you ask for an academic reference, and the AI knows from millions of examples what such a reference looks like … there’s a chance it’ll simply make one up. (Known as an “AI hallucination”.) To fix this, ask AI to give you actual links to evidence on the web – sources you can verify yourself. Here’s the prompt model: 

SAMPLE PROMPT: I am looking for published evidence that supports [assertion]. Provide links to such evidence, limiting your search to sources that contain actual research data and/or are on the websites of reputable journals, established institutions, experts in the field, or trade media.

All these techniques are simple and fast to apply – and they’ll make your writing rock-solid. Now let’s get a little more technical.

The advanced stuff: two bonus techniques

Last, a couple of bonus ideas. These last two are mainly for experienced users able to interact with the AI’s API directly, using a coding language like Python. But in a pinch you can still use the consumer app.

Bonus technique #1: Give yourself a crash course in a new subject

However much you know about a subject, your client is likely to know more – which means your first job is to learn about their product or service in depth. And AI can teach you. But building a personalised course takes more than one prompt – you’ll need a series of them, called a prompt application.

The method: first you ask the AI for a course outline (asking for revisions as needed) then go deeper into each part of that outline, to build lesson plans and later actual course content. Here’s a prompt to get started.

SAMPLE PROMPT: I want to learn about [X]. Give me a course outline that teaches this subject at [beginner, intermediate, expert] level, list all lessons in the course in logical sequence grouped into appropriate units. Give each lesson and unit a meaningful title.

Once you’ve got your course outline in the context window, ask for unit overviews and lesson plans in turn, building your initial list of lessons into a complete set of course material. Pro tip: as you go through, ask the AI to save each output as a downloadable file. Of course, all this work can be automated if you’re comfortable using the API most AI vendors offer.

Bonus technique #8: use custom instructions to bend AI to your style

Did you know you can set special conditions for your AI to follow? In ChatGPT (the most popular LLM) they’re called “Custom Instructions”, and you’ll find them in “Customize ChatGPT” by clicking your profile link in the app.

CUSTOM INSTRUCTION EXAMPLE: At all times use [American English]. Use language suitable for a reading age of 12 and an international audience; never use culturally specific jokes or puns. Accompany responses with links to supporting data wherever possible. 

You can use your customisations to do anything you want: give your AI an offbeat sense of humour, or speak in Klingon, or include an emoji in every sentence. As always, practice and experiment – you’ll be more productive in no time.

Conclusion: AI is a toolbox of techniques that adds value everywhere

People thought Amazon’s Kindle would kill traditional books; in fact sales of printed books are higher than ever. As streaming music has boomed, attendance at real-world concerts has risen. And of course machines like the Jacquard Loom made Northern England an economic powerhouse that created millions of jobs. 

In other words, new technologies tend to transform and expand markets – not destroy them. In the US$400bn content marketing industry, it’s likely AI will do the same. So take advantage of the opportunities it offers, with the techniques above!

Of course, the more advanced techniques take time and effort – but relax, Contentoo is ready to help you apply them at scale. You can book a free content strategy call whenever you are ready to explore new content solutions ;)

‍