How to turn a recording into a fully published post

The biggest reason you put off publishing is not a lack of ideas. It is that no one asks you questions to answer. When you sit and type alone, your ideas get stuck easily. But when you have the right questions, you will talk in one go without stopping.

Just follow the Know - Trust - Love funnel, and you will know exactly what to say first and what to say next. Add AI to sort your outline in seconds, and it takes you only about an hour of recording and 30 minutes of editing to put a finished post on your site.

Get your tools and questions ready

Before you hit record, you need three things: a microphone, a recording app and a set of guiding questions. If you record for the Write step, the locked outline takes the place of the questions: pick up the outline and talk.

Install a recording app on your phone

Your phone always has a microphone, so it is the easiest place to start. A laptop has a built-in one too. A desktop computer needs a separate microphone.

There are many recording apps. I use Tana Voice, because it records up to 2 hours and does three jobs for me:

• No lost files. With Wi-Fi or 4G, the recording goes up to a server for safe keeping.
• Text on its own. Once the file is on the server, your voice is turned into text on its own.
• Extra features. The app pulls out the highlights while you record.

I take my phone everywhere, and I can record as soon as I open the app. When I finish a recording of about 30 minutes, the file goes up to the server, turns into text, then syncs to my computer.

The phone is only for recording. Everything after that I do on the computer, because it is faster and easier.

So install the app on both your phone and your computer, and sign in with the same account, so the files sync on their own.

Write a set of questions for your post type

A set of guiding questions keeps you on point and stops you from repeating yourself. When you record, it is easy to repeat yourself: you cover an idea early on, forget it about 20 minutes later, and say it again. With a set of questions, you just answer them one by one, and your answers become the post.

Each post type needs its own set of questions:

Post type	The questions focus on	My number of questions
Know	Explaining so the customer understands	About 16
Trust	Step by step guidance, light on explaining	26 (this post)
Love	Strategy, process, opportunity and meaning: why you must do it, what it means for your customers and for you	36

Because the set is sized to the post type, your recording comes already split into blocks of ideas. Do not push yourself to answer every question. Just answer honestly what you truly understand.

1. Write one line of topic. It is a title, or a sentence you want to dig into. This line does not need to be very clear.
2. Pick the post type. Your post is a Know post, a Trust post or a Love post. The post type decides what the questions focus on and how many there are, as the table above shows.
3. Ask AI to write the questions. Paste the topic and the post type into the prompt below, then send it to the AI tool you use.
4. Keep the questions in front of you. While you record, you always know what you are talking about, and you do not repeat yourself.

The prompt I use to write the questions:

Play a busy, hard-to-please customer of mine. Write a set of questions for me to read aloud and answer myself while I record.

Topic: [one line of topic]
Post type: [Know / Trust / Love]
Reader: [who they are, what problem they have]

Rules:
- Make three groups, in this order: I. What it is, II. How it works, III. Why it matters and what you get if it works.
- The group that matches the post type goes deepest and holds at least 60% of the questions: Know is group I, Trust is group II, Love is group III. Ask each main idea as a chain of two or three questions, each one deeper than the last.
- Number of questions: about 16 for a Know post, 26 for a Trust post, 36 for a Love post.
- Each question is short, direct, one idea, 10 to 20 words. Under each one, add one short line in the reader's voice with one concrete detail.
- Ask at least 4 questions about my own story: the first time, a mistake, a turning point, a view of my own. Keep them open, and never guess the answer.
- Only ask. Do not answer for me, and do not make up my experience.

Remember three root questions when you have no AI

With no AI and no internet, you can still start recording any time with three root questions:

1. What is it?
2. How does it work?
3. What does it mean for you or for your customers, and why look into it?

When you answer these three, you will not stop there: your ideas slowly widen, and new ones keep coming. The good thing about recording is that it keeps your brain working, so your ideas never pile up.

Step 1 (Idea): Pour out raw ideas and lock the outline

In the Idea step, you say every idea you have about the topic, then let AI sort it into a sharp, lean outline. Instead of writing, you say it all, so the ideas come faster. AI pulls apart, orders and groups what you said, and you can even chat with your own words.

In front of a blank page, you have no ideas to make an outline, no outline to write from, and no words to edit. Recording gives you the raw material first. Recording shows up in two steps: the Idea step (say every idea out loud) and the Write step (talk instead of type). The Edit step is where you read and approve.

The general steps for this part are on the page make an outline from one line of topic.

Turn off the internet and answer each question

Recording is when you dig out what only you know, so you do not need the internet.

1. Turn off the internet and calls. Give 100% of your focus to the recording, with no internet, calls or anything else to break it.
2. Read each question aloud, then answer it. Skip any question you do not know. For a question you know and understand, say it all: numbers, memories, everything related. Where you do not remember well, say you do not remember well, because you need the truth.
3. Turn the internet back on when you finish. Only the recording app needs the internet, to send the file off to become text.

Talk as much as you can, and as exactly as you can. Recording puts a good kind of pressure on you: what you cannot say out loud, other people cannot understand either. Only when you can say it can other people understand it.

Not every recording turns into a post. After some practice, each of your answers will have depth and hold many ideas. My recordings break one question down into lots of small points, because I say everything I know and hide nothing.

Turn your voice into text with AI

By the time the file syncs to your computer, Tana Voice has already turned it into text. I am very happy with how well the app turns speech into text.

In the past, a mistake in a recording was very hard to fix. Now there are AI models made just for recordings. You talk, and the model:

• Drops the “uh” and “um”
• Fixes spelling
• Fixes words you did not say clearly
• Splits the text into paragraphs

The result is about 95 to 99% right.

I also tried Gemini’s speech to text model in Antigravity: Gemini works very well, as good as Tana Voice or better. Even so, I still choose Tana Voice.

Open the transcript on your computer and check the names: names of people, products and tools. For the rest, once I see the AI got it right, I do not check any more. You can still check a 17,000 word transcript if you want to, but usually there is no need.

Sort the ideas with AI to lock the outline

Give the AI the transcript together with your core message, so it pulls out the ideas that serve that message.

This is where AI helps most. A 17,000 word transcript takes the AI about 10 seconds. Doing it yourself takes 1 to 2 hours: reading it again from the start, splitting paragraphs, sorting each idea, then merging the ones that repeat.

The core message is just one sentence. This sentence is like a title, but it says more.

1. Write your core message in one sentence. That sentence tells the AI who the post is for, what problem it solves, with what solution, and what result. How to write it is in the guide create a core message for a post.
2. Ask AI to sort the ideas. Paste the transcript, the core message and the post type into the prompt below, then send it to the AI.
3. Sort again with your gut. Read each idea the AI gives back. Cut ideas that are extra, off the point or a bit too harsh, and add ideas you see are missing. If an idea is not clear, you can ask the AI back: “How did I explain this part in the recording?”, so you talk straight with your own material.
4. Lock the names of the main ideas. The names must be short, easy to understand and to the point, so anyone gets them right away.

The prompt I use to sort the ideas:

Below is the transcript of my recording. Pull out the ideas to make an outline for a post.

Core message: [one sentence: who it helps, what problem it solves, with what solution, what result]
Post type: [Know / Trust / Love]

Rules:
- Keep only the ideas that serve the core message. Merge ideas that repeat, and keep the strongest way of saying them.
- Give each idea a short name and a short paragraph that explains it, using the exact stories, examples, numbers and views in the transcript.
- Order the ideas by post type: for Know, explain so the reader understands. For Trust, follow the steps of the job. For Love, follow the strategy and the reasons.
- Do not add ideas, numbers or stories that are not in the transcript. Where something is missing, say clearly that it is missing.
- At the end, list the ideas you did not use, so I can keep them for other posts.

Transcript:
[paste the transcript of the recording]

The outline the AI gives back is not just bullet points: each idea comes with a short paragraph. Reading one idea with one paragraph, you see right away how that idea will be explained, which is much easier than reading small loose bits.

Do not be afraid to lose words. In the Idea step, I usually pour out about 50 ideas. Sorted by the core message, about 7 ideas stay. Keep the rest for other posts. This post needs to stay lean with 7 ideas, so keep only 7.

When you pick the ideas yourself without a core message, it is easy to catch the urge to keep every word: you say too much, then cutting 60 to 70% feels like a loss, so you stuff it back into the post. A stuffed post tires readers and scatters their focus. Readers need accuracy, focus and a smooth read, not lots of words.

Step 2 (Write): Record from the outline

In the Write step, you pick up the locked outline and talk instead of type. Talking your way into text is still writing, and the two ways differ only in speed. The outline has only about 6 to 8 ideas, so you can say them quickly and with depth.

I record because talking is much faster than typing. The recording for my Idea - Write - Edit post ran 85 minutes and came to about 17,000 words. In the same 85 minutes, typing would give me about 2,000 words. As I see it, talking is about 10 times faster than typing, and no tool today helps you type faster.

Skip the second recording if the first one went deep enough

Not everyone needs a second recording. If you already said everything you know the first time, a second recording usually adds only about 20%, which is not worth much.

I usually record only once and do the Idea and Write steps together. I write a detailed set of questions, then answer each one slowly and calmly: I dig deep, think, and say it all in that one recording.

I give it an hour, and that one hour finishes both the Idea step and the Write step. Skipping one recording already makes you twice as fast.

Task	My time
Recording for the Idea and Write steps	30 minutes to 1 hour
Polishing the outline into a post frame	30 minutes to 1 hour
Editing the post	About 30 minutes, about 15 with help from AI

Talk a lot, and skip the opening

When you record, you always say more than you need. The recording for my Idea - Write - Edit post had 17,000 words, and it was cut down to 5,400 words when published, only about one third. A recording always loses from 50% to two thirds.

So open the locked outline, say each main idea in order, go straight to the main content, and say more than you need about each idea, even when you repeat yourself or stumble.

Everyone talks at a different speed and has a different number of ideas in their head. People used to writing often talk less, so moving from writing to talking is not simple for anyone. You think faster and react better only through practice. There is no other way.

Repeating yourself, saying “uh” or “um”, or saying something wrong and then saying it again is normal. You cannot delete it the way you do when typing, but AI scans the text and deletes it for you. In the past this was a big problem. Now it is not.

Because you cannot delete it, recording makes you take more care with your words: you try to say things right, say more, and say any idea the moment it pops into your head. When people write, many have the habit of writing and deleting, and deleting one line 10 times is normal. If you already type three times slower than others, and you delete each sentence 10 times, you are 30 times slower.

A recording has no opening, and I do not try to make one. The questions do not aim at an opening. You go straight to the main content, without circling around to think of ways to hook or keep the reader.

Once you have a core message, an outline, and answers to all the questions, the opening comes on its own. The opening and the ending are added in the Edit step.

Look into where you got stuck, then record again

Getting stuck on an idea of the post while you talk is the sign you need a second recording. The first time, you only went shallow: you said what you remembered and did not dig deep. Play the customer and ask yourself: if a customer asked you that exact question and you could not answer, you are stuck.

Getting stuck on an idea that has nothing to do with the post is fine. You do not need a second recording. You need to record again when you get stuck on an idea of that post, for example:

• An idea you cannot explain.
• A method whose name you cannot remember.
• A process where you only remember step one and forget steps two and three.

1. Say everything you can say. Do not stop where you got stuck: say everything you can say first, then note the places where you got stuck that matter to the post.
2. Look into the gaps. Read your documents, your notes, and trusted guides by other people again.
3. Understand it again in your own way. Do not copy, and do not paste. Once you fill the gaps, you can explain to customers with the most familiar words, so they understand fastest.
4. Record those ideas a second time. Record again from your own understanding, until you can talk in one go.

The general steps for this part are on the page write a draft.

Recording is much easier when you already understand. That is why I often bring up the Capture and Share challenge: capture means learning something new deeply, in your own understanding, and share means breaking what you just learned into small parts yourself, without asking AI to do it (sharing by recording works too).

When you break it down yourself, you remember it. With a daily habit of capturing and sharing, the ideas pour out when you record. Without that habit, you trip easily: you think you are good in that field, but when you record you cannot remember a single word, and you cannot answer when a customer asks.

Step 3 (Edit): Approve the draft and publish

In the Edit step, your job is to read and approve. Recording only speeds up the first part of the process. The reading and approving are still yours.

Approve each idea the AI ordered

AI only edits your draft. It does not write whatever it likes. Unless you ask it to copy someone’s voice or style, its job is only to reorder, split and fix the parts that are not clear. The transcript is the source. The AI edits from it, so it does not dare to change things wildly, even though every AI makes things up at times.

1. Ask the AI to order the draft by the outline. Tell it to only order, split and clarify, and to copy no one’s voice or style.
2. Read each idea, one by one, without slacking. Fix the places where the AI did not get what you meant.
3. Ask the AI to present hard ideas again. If an idea is too long or too hard, ask the AI to present it more clearly. The idea stays the same. Only the way it is presented changes.

A sample prompt to ask AI to order your draft:

Below are my locked outline and my raw recorded draft. Reorder the paragraphs to follow each idea in the outline exactly.

Rules:
- Keep the wording, the ideas and the examples of the original draft.
- Only fix clumsy phrasing, cut filler words ("uh", "um", repeated words) and split the text into clear paragraphs.
- Never make up new ideas or change my style.

Outline:
[paste the locked outline]

Draft:
[paste the recorded draft]

You approve last, not the AI. If you see a mistake and leave it, that is your fault, not the tool’s. I read each block closely: as many ideas as there are, that many I read. I never hand it all to AI and hit publish.

Add the opening, ending and description

The opening and the ending are added in the Edit step, because you went straight to the main content when you recorded. The title usually stays the same, because it was locked in the Idea step with the main ideas and the smaller ones.

Write the opening from your core message and what you answered, then write the ending. After that, make the description clearer, and adjust the SEO title if needed.

The SEO title has a length limit. Sometimes your title is just right but one or two words too long, and the SEO title gives you the chance to adjust it. The same goes for the description.

Publish first, add pictures later

On any website platform, copy and paste is the fastest way to publish. If you have an AI assistant, ask it to publish for you, for example through MCP. Without an AI assistant, you do each task yourself.

1. Paste the post into your website. Copy the post and paste it into your website platform, or ask your AI assistant to publish it for you.
2. Finish it, then hit publish. Add the cover image and any pictures you have, format the text, add tables and line breaks. Check the SEO title and the description, then hit publish.
3. Note what is still missing. Wherever a picture or video is missing, note it, then come back later to finish it.

A how-to post often needs pictures and videos: screenshots that help readers follow your real experience. If you have them, add them right away in the Edit step, then publish. If not, publish first, finish later, so the post does not sit on your computer too long.

Once a post is out, it is your job to finish it. A post that sits on your computer is easy to put off. Build this habit: publish first, and if a video or pictures are missing, come back and add them later.

Start your first recording

Your recording holds your own knowledge, experience, views and memories. AI cannot get inside your head to pull those out. So a post born from your voice is 100% your own: nothing reheated, nothing borrowed from anyone. That is the only way for your content to beat AI.

Finish one post a day, and in 3 months you have 90 posts that are all your own. The three steps in this post close one real round of the Idea - Write - Edit process.

Today, pick one line of topic, open your questions, and hit record.