NEW Introducing the Pro Dojo. Become more efficient with your favorite AI writing tools with exclusive perks. Become a pro ninja here.

How to Use AI to Write Efficiently: Context in ShortlyAI

Context is the content used to guide the AI to return the results you want. Understanding this context could be the most important skill you'll need to become efficient with ShortlyAI.

This guide will explain everything you need to know to make the most of your AI writing assistant and become more productive than you ever imagined you could be.

Before we dive in, let's go over the Shortly UI's various parts where context is important.

The ShortlyAI UI

The truly amazing thing about Shortly is that it is both straightforward but subtly complex at the same time. On the surface, there aren't many things to worry about. You can simply write your heart out.

And when you need AI assistance, it's there. And the one thing with total control over it is your words -- and how you use them in the article brief, title, content area, and the special /commands.

This is something unique to Shortly. No other AI writing app gives you this level of control and flexibility.

Let's go over the article brief and title first.

The Article Brief & Title

The article brief tells your AI assistant what type of content you want and the topic it is about. Providing a detailed, topic-rich brief is important.

For example, "Please write an article about working from home" is not a very topic-rich article brief.

Instead, use something like "Please write an article about the difficulties of dealing with distractions while working from home."

You could go further even "Please write a creative article about the difficulties of dealing with distractions while working from home with young children."

This is important because the article brief is the "anchor" of the context sent to the AI. When you click the "write for me" button, it is always sent (along with the title) -- so it has a lot of influence over the AI's result. I'll explain more about why I emphasized the word "always" here in a moment.

The title acts in the same way, and the best practice I've found is to use it to set the overall topic. In the example above, a good title might be "Working From Home: Overcoming Distractions From Your Little Ones."

Notice I used slightly different words (with the same meaning) -- this will set the stage for the article content area.

Before we move on, though, I want to emphasize the article brief and title are almost always sent to the AI. A little later I'll explain when it isn't used and why.

The Content Area

This is where things start to become interesting. Now the article brief, title, and the content you’ve begun writing affect your AI results when you hit that “write for me” button.

But there are a few things to keep in mind about the content area.

First, only the content above the line where your cursor is when you hit “write for me” is used. Nothing below that line is used at all. You could be on paragraph 2 of a 30 paragraph article and none of the following 28 paragraphs would have an affect on the output.

Next, you need to be aware of patterns.

Patterns

The GPT-3 AI LOVES patterns. If you start a bullet list, the AI will return bullets. This can be both a benefit and a detriment.

On the one hand, you can harness this to enable some exciting scenarios (like this AI inception video where I show you how to get the AI to create AI commands). For instance, if you are looking for headline ideas for your topic, you'd want the AI to follow the list pattern you start (or better yet, get from the /instruct command).

Other times you might become trapped in a pattern when you don’t want to be — particularly when writing long-form content. The way to recognize if you're trapped is when the AI keeps returning content you don't expect, but seems to match the content above your current location in the document.

You’ll learn some ways to deal with this in a moment. Next, we need to go over how context works in the /commands.

Context in the Refine Commands

The refine commands are the shorten, rewrite, and expand commands. I call them the “refine commands” because they are used on existing text to make refinements to them.

The refine commands have a very special context. ONLY the content within the [] of the command is used — nothing else. No content in the article, no article brief, no title. Just the [text].

For this reason, you need to be careful. For instance, if you use /expand [best parks], the only thing the AI has to work with is “best parks.” Imagine if your boss gave you a writing assignment to write about “best parks.” You may have some difficulty.

Be mindful of this and add some topic-rich content — something like [best public parks with a view in NYC]. This will give the AI something to work with and likely result in the content you can use.

Now let’s move to the instruct command as it has its own special context too.

Context in the Instruct Command

This is where things really become interesting because the instruct command is so special in Shortly. It’s literally called instruct because you can imagine as if you’re turning to your AI assistant and giving it instructions to get the content you want.

And the real power behind it is in how the context works. For the instruct command, only the content above the command and within the [] is used. The instruct command does NOT use the article brief and title.

Can you imagine what’s so special about this?

No?

What about something like /instruct [give me some questions about the above]? Does that sound interesting?

Maybe /instruct [write a conclusion to the above]?

This context opens up the command to give you a ton of flexibility. Explore and experiment to find new ways of using it. Now that you know how all these various parts of Shorty use context… you’ll need a way to control it sometimes.

Control: Context Cutoff Ruler

Although I don’t think Shortly calls this the “context cutoff ruler,” I do. It helps make sense of what it does.

It literally cuts off the context at the point where you type ///.

This is useful for a few reasons.

One, if you find yourself trapped by a pattern. For instance, if the content above where you’re writing is a list, but you no longer wish to write in a list format — the /// can help.

Two, if the topic of the content above the current point where you’re writing is different than what you want to write about, the /// can help.

And finally, you can also use the cutoff ruler when you want to isolate a /instruct command from the content above it:

///
/instruct [please write some persuasive bullets about Birdie, my revolutionary new backyard bird feeder…]

Alright, now that you know what all the various context points are and how you can control them, is one type of context more important than another?

Weights of article brief and title versus content

Right away, I want to say that none of the context has more weight than another. It’s all the same. They all work in concert with each other.

But there is something to know when your article becomes very long. The article brief and title are ALWAYS sent when you click “write for me.” But if you recall, so is the content above the current point where you’re writing.

But there is a catch. Because there is a limit to how much context can be sent to the AI, the content above your current position might get truncated.

For instance, if you were 2,000 words into a long article and clicked write for me — the article brief, title, and a large amount of the content will be used — but the top will be “shaved” to fit the limit.

If that makes sense.

So in a small way, the article brief and title have a tiny bit more importance because they’re always sent — but do not have any extra weight in what will be returned.

With that in mind, there’s another important thing to know about the article brief and title.

Change the article brief and title as needed

Since the article brief and title are definitely used when the “write for me” button is used — you can use them to tweak the context as you work through your article.

Perhaps you’d start with a more general article brief like “Please write about the best parks with a view in NYC,” but update it later in your article when you’re writing about something specific.

The main takeaway here is that you can change the article brief and title — they’re not static.

Don't forget the power of expand

This is almost like a secret weapon. When you’re writing, and you’re deep into a long article — and need a quick definition on something, or content completely outside the scope of the content around it — there are two ways you can do this.

One is with the power of /expand. If you recall, it’s one of the refine commands, and so the only context used is what is between the [text] of the command.

So it doesn’t matter what is around it. Nothing will interfere — /expand [the definition of sushi] should give you the definition of sushi, even if right above it there is something else about sushi.

The other way you could do this would be by combining the /// cutoff ruler with /instruct.

///
/instruct [please define sushi]

Since the cutoff ruler will stop the instruct command from using the content above it — it should write out the definition of sushi here.

Alright, that’s the context in Shortly… just one more thing that will help you create the best content.

Fine-Tune Your Content with the Granular Controls

I have a whole video dedicated to these controls, so I’ll go over them quickly. Within the /instruct command, you have the ability to fine-tune what the AI will produce.

You do that by using ++,+++, and --, --- before the keyword, you either want in the output or don’t.

A +++ means you really want it in the output, a --- means it will not be used.

For instance, /instruct [please define sushi +++Japanese ---wasabi] — this means you really want the word Japanese used in the output, but wasabi will not appear.

Use these when the AI is close to what you want but isn’t quite perfectly in tune with your expectations.

Phew… alright, now you have all the tools you need.

Now you Dance and Ask

With what you just learned about the context, now it’s time to write — and in ShortlyAI, that means two things: dancing and asking.

I know… sounds strange, but hear me out.

What I mean by the “dance” is you write a little, maybe a sentence and a half, then hit the “write for me” button to let the AI finish your thought. You do this back and forth until your article is complete — this is the dance.

Every once in a while you want something more specific. This is the “ask.” You use the /instruct command to ask your AI partner for a specific type of content.

And essentially, to write long-form content — whether that’s a 4,000-word blog post for your website or a 70,000-word book — you dance and ask your way to the finished work.

Summary

That's a wrap. I hope you find this guide helpful. Go over this as many times as it takes for you to really understand how context works in Shortly. This is vital for getting exactly the content you want from the AI and working efficiently.

Good luck and happy writing!