How to balance six platforms? What you need is not writing, but an efficient production line

✍️ This article was originally published in “Vista Newsletter” No. 131, originally published in iamvista.substack.com (opens in a new tab). If you like this kind of content, welcome to free subscription to “Vista Newsletter” (opens in a new tab) to receive my observations and methods first-hand every week.

Hi, I’m Vista.

A while ago, a friend came to me for advice on building a personal brand.

This friend named Amy said that she knows the truth of “content is king” and understands the importance of regular output. The problem is that she is too busy and spends a lot of time writing articles every time.

So, I asked Amy to do an interesting thing: get a code sheet and record all the steps she took to produce an article.

It’s not just the journey of sitting down and typing, it’s the whole thing. From imagining the topic, checking information, cross-referencing the data, writing the first draft, settling down to revising the draft, and then rewriting the same article into a blog version, a Facebook version, a LinkedIn version, an X version, a Threads version and an e-newsletter version.

Amy told me an incredible thing, adding up all the time before and after, it actually took 11 hours.

Amy later told me that what really surprised her was not that it took so long, but that it took almost the same amount of time to write an article every time. She has been writing for self-media for several years, and she has indeed accumulated experience, but her speed has not kept up: today’s article takes about the same amount of effort as the one ten years ago. This means that every article she writes is an independent handicraft that falls apart as soon as it is finished, leaving nothing behind to make the next article faster.

So, in this newsletter, I want to tell you the complete story about how I taught my friend Amy to deal with this matter. At the end of the article there is a map I compiled and can be given to everyone for free! You can just take the pictures and build it yourself, there is no need to take lessons from me.

What ate up 11 hours?

Well, let’s get this settled first, because I find that most people are saving the wrong place.

Most people think that what blocks them is “writing”, so they spend all their time writing, or try to find a better writing prompt. But if you break down the process, it is not difficult to find that writing the first draft actually only takes about two hours, and the remaining nine hours are distributed in these areas:

• Research and verification: looking for information, reading reports, and confirming whether the numbers are quoted correctly, it will take one to two hours to get rid of it.
• Revising the draft and removing the AI ​​flavor: The first draft given by AI is usually correct but empty. It reads like boring Wikipedia, and you have to change it sentence by sentence back to what you would say.
• Platform-by-platform rewrite: This is the biggest black hole. Five platforms, five word counts, five tones, and five ways to start. If you grab 40 minutes on each platform, it adds up to more than three hours.

Well, do you see it? What really eats up time are those highly repetitive processes that start from scratch every time. They don’t require your talent, just your hours. And you just tied the most expensive things (your judgment, your opinions) and the cheapest things (typesetting, rewriting, and finding information) to pay within the same period of time.

Therefore, my conclusion is: the problem is not the writing style, but that many people have been using the method of working on a single project to do something that should be done by the production line.

The difference between handicrafts and production lines lies in whether there is anything left over from the last time

Okay, let’s use an analogy. The so-called craftsmanship means that you start carving from a piece of log every time. No matter how well you do it, the hundredth one takes the same effort as the first one. But the production line is different. The focus of the production line is not to run fast, but to have clear inputs and outputs for each process. So if you adjust a certain process this time, it will look like the adjusted next time.

Improvements will settle down, and this is the real value of the production line.

I later broke down my entire set of content production into five layers. These five layers are not the structure I came up with while sitting at my desk. They are the shape that remains after I dismantled and classified those 11 hours piece by piece, and then asked myself “What is the input of this section, what is the output, and can it be handed over to the machine?”

Five-layer content production line

The first layer: input layer (one sentence → seed card)

Inspiration (opens in a new tab) most often dies before being written down. You come up with a great idea on the subway, tell yourself to write it down later, and then it disappears.

The input layer has only one thing to solve: lowering the recording threshold to close to zero. My approach is to use voice recording, say a paragraph about what comes to mind, and let the system organize it into a structured seed card: what is the core proposition, who is it written for, possible corners, and what is my current mood.

The point is, what comes out is structured. Three months later, when you look back at those scattered inspiration notes, you will not remember what you are talking about. On the other hand, if you look back at a seed card for three months, you can still continue working directly.

Second layer: Research layer (Seed Card → Research Brief)

I use a five-step research method at this level: first define what I want to check, then search extensively, cross-reference, mark the credibility of each fact, and finally compile it into a research brief.

The most critical action here is to mark credibility. AI can find information very quickly, but it will tell you the “general opinion in the industry” and the “measured data of a certain study” on the same plane, with the same firm tone. Therefore, I asked each fact to be marked with its source and strength. When I wrote it, I knew which sentence should be categorical and which sentence should be added with provisos.

The one to two hour process is compressed into the time it takes you to make a cup of coffee and come back.

The third layer: production layer (research brief → first draft of the article)

Only here is it’s turn to “write”. Whether or not you write like you depends on two things: your style profile and narrative framework.

The narrative frame determines the skeleton of the article. On the same topic, if you write using problem-based solutions, using a control group, using a timeline, or using personal stories to bring out insights, they will be four completely different articles. Most people don’t realize that they are unconsciously using the same frame every time, so the article all reads the same.

The fourth layer: processing layer (first draft → interesting article)

This layer is the layer that is most easily skipped but cannot be omitted in the entire production line.

No matter how neat your first draft is, it still smells like AI: there are too many adjectives, the sentences are unnaturally long, and you say a little bit of everything but not one sentence that only you can say. The processing layer does three things: removes the tone, injects real story and details back, and calibrates the tone.

A student gave feedback after taking the class. I think he taught it better than me:

What impressed me most was not the five automated processes, but the very first step: building your style profile. I threw in the articles I wrote in the past, and 10 minutes later it told me: you prefer contrasting structures, you like specific numbers, and you like to ask the reader a question at the end. To be honest, it’s more accurate than what I can say myself.

Did you see it? This is what style files do. You are not teaching AI to imitate a style. You are asking it to find the style you already have and write it into specifications, and then every article will be aligned according to this specification.

The fifth layer: Distribution layer (refined article → six platform versions)

There is only one sentence to remember at the distribution layer: **Reinterpretation, not truncation. **

Chop a 3,000-word article into 200 words and throw it on Threads. That’s not distribution, that’s just wreckage. Threads wants a hook that can make people stop their fingers; LinkedIn wants professional judgment and transferable methods; Facebook wants situations and stories; newsletters need to be complete and have a sense of conversation.

Same core message, six different ways of saying it. It takes 40 minutes for a person to do this on one platform; with the comparison table and specifications, the machine can produce six versions first, and you only need to spend time polishing and setting the tone.

What this production line saves everyone in the end is actually not time.

After the run-through, one article went from 11 hours to about 30 minutes. But what really changes the way we work is not the ten and a half hours saved.

It allows everyone to finally focus on the part that only they can do.

For example, when it comes to research, compilation, and format conversion, machines can do it more stably than people, and they won’t be tiring. As for opinions, judgments, whether to call someone for a detail, and which words will hurt the readers, I will always have to make these decisions myself. In the past, these two kinds of work were mixed into the same 11-hour period, and I was so tired that I often made the judgments that required the most energy when I was least energetic.

Now the order is reversed.

There is another side effect that I did not expect: after the output stabilizes, publishing articles no longer requires courage. Six articles a week, 312 articles a year, sounds scary, but when it’s a running production line rather than a ritual that you have to start over every time, you just have to keep moving forward.

Something you can start today

If you read this and feel that the entire system architecture feels huge, I will give you a starting point that can be completed tonight.

Don’t start building from the first floor, start from the style section between the third and fourth floors.

The method is simple: post all the five to ten most satisfying articles you have written in the past to AI, and then ask it to analyze your writing characteristics, including usual sentence length, punctuation habits, commonly used metaphor types, your preferred opening method, and which words you avoid. Ask it to be output as a spec document that can be quoted again and again, not as a complimentary review.

Within half an hour, you’ll have a copy of your writing specification. This thing alone will make every first draft you produce with AI a lot closer to your tone.

This step doesn’t require any additional tools or money. Yes, you can do it tonight after work.

That picture is given to you for free

I drew the entire production line into an “AI Content Production System Blueprint” and put it on the website for everyone to receive for free:

👉 Get “AI Content Production System Blueprint” for free now

There are four things in it:

• A complete map of the five-layer content production line (input, output and handover points of each layer)
• Guide to the selection of four narrative frameworks (which framework should be used for what subject matter)
• Three-step refining method to remove AI cavities
• Six platform distribution comparison table (specifications of word count, tone, opening method)

Fill in your email and the email will arrive within one to two minutes (remember to check your spam box). The download link will be valid for 24 hours.

One thing I want to make clear first: **This diagram is designed to let you build it yourself. **The five layers of logic, the input and output of each layer, and why it is cut like this are all written on the diagram. You could just take the picture and spend a few weekends putting it together yourself. In fact I’d be happy to see someone do that, because that would make the picture clear enough.

If you want someone to do it with you

Of course, there is still a gap between getting the map and actually doing it once. How to install the tools, how to adjust the style file so that it really looks like you, and how to write the specifications for the six platforms, these details can only be known by doing it.

Therefore, I held a three-hour practical class to install the entire production line into your own computer on the spot, including installing the toolkit, creating your style file, and then running from one inspiration to six platform versions, so that you can continue to use it when you go home.

“AI Content Production System Workshop”: August 30, 2026 (Sunday) 09:00 to 12:00, Taipei City, limited to 16 people, please bring a laptop.

Course information is here: vista.tw/courses/ai-content

It’s up to you to decide whether you want to come or not. You are welcome to take the picture away for free and try it yourself. I think this is a good choice.

##Finally

I have often thought recently that the real difference that AI brings to content workers may not lie in whose prompts are written more beautifully. What matters is who is willing to break down their profession into processes that can be performed repeatedly.

Artistic creation naturally requires careful craftsmanship. But if it is business writing, it is still important to reduce costs and increase efficiency. Because as long as it is still a handicraft, your output will always be equal to your working hours; once it becomes a production line, your experience will start to accrue.

This e-newsletter is here and is about to come to an end. If you only remember one sentence, I hope it is this sentence:

**What you lack is never writing skills, but a production line that allows good practices to be retained. **

See you in the next e-newsletter!

Vista