Captions that sell,
written while you sleep.
For years I typed every caption by hand, posted decent content with weak copy underneath it, and watched the engagement go nowhere. Now a small AI reads each image or video, writes the caption, adds the call to action, and posts it. Around 250 captions a month, without me touching the keyboard. Here is exactly how it works, and the prompts you can use to build your own.
The content was already good. The copy underneath it was not.
Most creators spend real energy finding or making great images and videos. Then they type a lazy caption, add a few hashtags, and wonder why no one clicks. This machine reads the visual content, decides what the caption should say, and writes one that actually sends people somewhere. Here is who it is for, what goes wrong without it, how it works, and what you get back.
Who it’s for
Creators, digital product sellers, affiliate marketers, solopreneurs, anyone who posts regularly on social media but spends far too long writing the copy underneath each post. If you have ever stared at a blank caption box for five minutes and then typed something forgettable, this is for you.
What goes wrong
Without a caption writer, the copy is always the weakest part of the post. The image attracts the eye but the caption loses the click. People scroll on, the product never gets seen, and the content that cost hours to curate earns almost nothing. Great visuals with weak copy are a wasted investment.
How the machine works
A small vision AI reads what is actually in the frame of each image or video. A second AI then uses that description to write three caption candidates, each one designed to direct the reader toward a product or article. The best option gets posted automatically. You never type a single word.
What you get back
Around 250 captions a month written and posted without you, each one pointing toward a product or an article that earns. Content that already works hard on the visual side now works equally hard on the copy side. Every post becomes a small, quiet salesperson.
The content was being curated. The captions were being typed by a tired brain at midnight.
For a long stretch I ran twelve Instagram accounts, each posting three or four times a day. The content side was handled, the curation was automated, and the videos looked great. But then each post needed a caption, and I was writing them by hand, or worse, pasting the same lazy description under twenty different posts.
The captions were doing almost nothing. Weak copy was sitting under strong visuals, and the people who stopped to look were not clicking anywhere. I was putting real energy into the content and leaving the part that actually sells completely unattended.
So I built a small chain. A vision model looks at each image or video before it posts, reads what is actually in the frame, and writes a short description of the content. A second AI takes that description and produces three caption candidates, each one closing with a soft direction toward my product or an article.
The machine picks the strongest of the three and posts it. The whole thing runs overnight. By the time I wake up, accounts have been posting sharp, sales-pointed captions since midnight. I typed nothing. I made no decisions. The copy under every post is doing its job.
Over a month that adds up to around 250 to 300 captions written and posted, each one trying to send a reader somewhere useful. The content does not just look good anymore. It works. And it all runs without me.
Proof point: I have documented how my accounts run on YouTube in $3k+ With One Social Post and across my Growth Hacking series, so the modest figures on this page are real and checkable.
See the frame, write the copy, close the click
What made this work was the vision step. A caption writer with no idea what is in the image produces generic copy. Feed it a real description of the frame first and the caption becomes specific, credible, and far more likely to earn a click. Three steps in order, and the post does its own selling.
See before you write
The first AI reads the image or video frame before a single word of copy is written. It identifies the subject, the mood, the setting, and the detail that will resonate. A caption written from a real description of the actual content is always more specific and more credible than one written blind. Skip this step and the copy is generic. Do this step and the copy sounds like it was written by someone who actually looked.
Write three, post one
The caption writer produces three candidates, not one. One version might be playful, one direct, one educational. The machine scores them against the goal set for that account, picks the strongest, and posts it. The two that are not picked are discarded automatically. Three drafts for the cost of zero effort means the live post is always the best version, not the first draft.
Every caption closes with a direction
A caption that ends with nothing is a caption that earns nothing. Every output from this machine closes with a soft, natural direction toward either a product page or an article that leads to one. The direction is not a hard pitch, it is a small door left open for the reader who is already interested. Over 250 posts a month, those small doors add up to real traffic and real sales.
See the frame, write three, post the best one with a direction. Do those three in order and every post quietly earns its keep. Skip any one and the content goes out looking good but doing nothing.
What captions used to look like
- A tired brain typing the same lazy description at midnight
- Generic copy that could go under any post from any account
- No call to action, no direction, no reason for anyone to click
- Hours of energy into the visual, nothing left for the copy
- Good content with weak copy that earns almost no traffic
What captions look like now
- A vision AI reading every frame before a word is written
- Three caption candidates written from the actual content of the post
- Every caption closing with a soft direction toward a product or article
- The whole chain running overnight, captions posted while you sleep
- Around 250 specific, sales-pointed captions posted per month, on autopilot
How I describe what is in each image or video
The whole chain starts here. A vision AI reads the frame before a caption is written. This is the prompt I use to make sure the description is specific enough to be useful, not just a vague summary that produces generic copy.
Read the frame and describe it precisely
I am going to give you an image or a short video. Look at it carefully and give me back a precise, useful description. Include: - The main subject (what or who is in the frame) - The setting or context (where it appears to be taken) - The mood or tone (bright and energetic, calm, professional, etc.) - One specific detail that stands out and could anchor a caption - The likely intent behind the post (educational, inspirational, product showcase, lifestyle, etc.) Keep the description under 100 words. Be specific, not vague. Write only what you can actually see. Do not invent context. This description will be used to write a caption, so every detail you include is a tool the caption writer can use. Output the description only, no extra commentary.
A precise description is the difference between a caption that sounds like it was written by someone who actually looked at the post and one that could go under anything. Spend the effort here once and the copy downstream earns it back every time.
How I write three caption candidates from that description
Once the frame is described, this is the prompt that turns it into three real caption options. Writing three and picking the best is the move that keeps the live post from ever being a first draft.
Write three caption candidates from the description
Using the description I just gave you, write three caption candidates for this social media post. Rules for all three: - Under 180 characters each - Written in a warm, first-person voice - Grounded in the specific content described (no vague generalities) - Each one ending with a natural, soft sentence that directs the reader toward an article or a product (I will give you the link or the destination separately) - No hard pitches, no phrases like 'buy now' or 'click here' Write the three versions with different angles: - Version A: Direct and clear, tells the reader exactly what the post is about - Version B: Leads with a question or a small story hook - Version C: Educational, gives one useful insight from the content Output the three versions, labelled A, B, and C.
Three versions means the best one goes live, not the first one you wrote at midnight. The extra two minutes this takes pays for itself in every post that earns a click instead of a scroll.
The 3-minute overview of how this works
Before the build steps, watch this short overview. It’s the exact video from the Automations Made Easy page, and it walks through the mechanics behind machines like this one. 1,000+ students have used these mechanics to save two hours a day, with zero coding.
Want to build a caption machine like this in your business?
1,000+ students. Save 2 hours a day. No coding required. In Automations Made Easy you’ll learn the simple skills to build little machines that read your content, write the copy, and post it on autopilot, just like this one.
What the caption machine quietly does every week
Let me keep this small and believable. I am not going to tell you it makes every post go viral. I will tell you what a modest version really does for someone who posts three times a day across multiple accounts.
Say you run three accounts posting three captions a day each. That is around 270 captions a month the machine writes instead of you. Even if only one in fifty posts earns a product click that converts, at a modest product price you are looking at steady, quiet income from content that would have gone out with lazy copy anyway.
Around 270 captions written per month, without typing a single one
Every caption pointing toward a product or article
Over a year, roughly 3,200 sales-pointed posts across your accounts
All without you sitting down to write copy for a single one
Now let it compound. 3,200 posts a year, each one with a real direction, is a quietly growing funnel that feeds itself. The ones that earn a click today keep earning from saved and shared versions for months. I documented my real results on YouTube, including $3k+ from one social post and my growth hacking series.
I wrote more about how I run my social media accounts without being on social media in my Substack, including how the caption quality keeps improving over time as the AI learns the accounts.
How I add the call to action toward a product or article
A caption without a direction is a caption that earns nothing. This prompt is what I use to layer the call to action onto the best caption candidate, keeping it soft enough to feel natural and specific enough to earn the click.
Add the soft call to action to the chosen caption
I have chosen one of the three caption candidates. I am going to give it to you now along with the destination I want people to visit (a product page or an article). Your job is to rewrite the ending of the caption so that it closes with a natural, soft direction toward that destination. Rules: - Do not add more than one sentence to what is already there - The direction should feel like a logical next step, not a pitch - If the destination is a product, frame it as something useful the reader can get, not something I am selling them - If the destination is an article, frame it as more detail or a deeper look at what the post hinted at - Keep the total caption under 220 characters Output the full, final caption ready to post. Nothing else.
Every post is now a small door left open for the reader who is already interested. Over 250 posts a month, those small doors add up to traffic and sales that the old lazy captions never earned.
How I rewrite a caption in a different tone
Sometimes the caption that works best for one account is wrong for another. This prompt lets me take any finished caption and rewrite it in a different tone, so the same content serves a professional account, a playful one, or an urgent one without me writing from scratch each time.
Rewrite the caption in a different tone
I am going to give you a finished caption and ask you to rewrite it in a specific tone. The core message and the call to action must stay the same. Only the voice and the angle change. The three tones I use most often: - Educational: calm, useful, focused on one clear takeaway. Reads like a short lesson. No hype. - Playful: light and informal, like a friend talking to another friend. One small joke or wink is fine. Nothing forced. - Urgent: direct and honest about why the timing matters. No false scarcity. Just a real reason to act now. I will tell you which tone I need. Rewrite the caption in that tone, keeping it under 220 characters, and output the new version only.
One caption, three accounts, three tones. The same content works for a professional audience in the morning and a playful one in the evening, with zero extra writing from you.
The exact build, step by step
Feed the bot a photo or video
One time connection setup. You point the machine at the folder or queue where your social content lives before it posts. Every image or video that arrives there automatically becomes the input for the caption chain. You never paste anything manually. The visual content is the trigger.
The AI reads what is actually in the frame
A small vision model looks at the content before a word of copy is written. It identifies the subject, the setting, the mood, and the one specific detail that a caption can anchor to. Without this step, the copy writer works blind and produces generic text. With it, every caption is grounded in what is actually in the post.
Decide the goal: sell, click, or opt in
Each account has a goal set once during setup. One account points to a product page. Another points to an article that leads to a product. A third points to an opt in page. The machine already knows which goal belongs to which account, so the caption it writes always closes with the right direction. You set it once and never touch it again.
Generate three caption candidates
The caption writer takes the frame description and the account goal and produces three versions. One direct, one with a story hook, one educational. Three different angles on the same post. The machine scores them and queues the strongest. You never see the other two.
Pick the strongest and post it
The machine scores the three candidates against the goal and the account’s tone, picks the best one, and posts it. No human review required. The caption lands under the post with a soft call to action already embedded. The post goes out complete, every time.
Track clicks back to the product
A small link tracker sits between the caption and the destination. Every click from every post is counted. After a week you can see which accounts, which post types, and which caption angles are earning the most clicks. The machine keeps improving quietly, post by post, without you having to guess what is working.
A photo or video enters the queue
The machine wakes up on its own and starts the caption chain in the background
A vision AI reads the frame
Subject, setting, mood, and the one anchor detail that makes the caption specific
Three captions are written and scored
One direct, one story hook, one educational, each closing with the right direction
The winner posts with the call to action
Clicks tracked, account fed, product pointed at, all without you typing anything
Want to build this in your own business?
Automations Made Easy teaches you, step by step, how to build small caption machines like the one you just read about. Zero coding. Plain English. 1,000+ students have already used it to post sharp copy on autopilot.
The six months after I switched it on
Here is the modest shape of it. The first few weeks are slow as the machine learns which caption angles earn clicks on each account. From week four the scores stabilise and the click rate across all posts settles into a steady rhythm.
Average click-through rate per post per week (modest estimate, 3 accounts)
Notice the line never spikes and never claims anything heroic. That is the point. A steady, growing click rate across hundreds of posts a month is a quiet engine that sends real traffic to real products, every day, without you sitting at a desk.
What other students built
I teach the simple skills behind machines like this in Automations Made Easy. Students who built their own version sent back what changed in their first month.
“I was posting great content with terrible captions. The bot reads each image and now every post closes with an actual reason to click. My link traffic doubled in the first month.”
“The vision step is the part that surprised me. It actually describes what is in the photo, so the caption sounds like it was written by someone who looked at the post. Not generic at all.”
“I run four accounts and I was spending two hours a day just on captions. Now I spend zero and the quality is better than what I was writing tired at midnight.”
“The three-candidate system is smart. The first draft is never the best one. Having three versions scored and the winner picked automatically made a real difference to engagement.”
What’s inside Automations Made Easy
AME isn’t a library of pre-built automations. Every business is slightly different. What’s reusable across all of them is the underlying mechanics: how to set up little machines that listen, write, and follow up while you sleep, and how to wire the pieces together without writing code.
The program walks you through six modules: The Right Tools (the cost-effective, no-code stack I actually use), Task Selection Mastery (which automations are worth building first), Design Secrets (mapping an automation before you build it), Zero to Hero (complete beginner to confident automator), Real-World Application (we build a full automation together, end to end), and Monetization Mastery (turn the skill into a side-business).
It also includes done-for-you templates you import in two clicks, over-the-shoulder training videos, and the same playbook 1,000+ students have used to save two hours a day. No coding required. If you can copy and paste, you can build this.
The copywriting bot: common questions
Pulled from what readers and Automations Made Easy students ask most.
Does the bot actually understand what is in my posts, or does it just write generic copy?
It reads the frame first. A vision model looks at the actual image or video before a single word of copy is written. It identifies the subject, the setting, the mood, and the specific detail that will anchor the caption. The copy writer then works from that real description, not from a blank brief. Generic captions come from bots that write blind. This one looks first.
What if the caption it picks is not quite right?
You can add a review step. Instead of posting automatically, the machine can queue the winning caption for your approval and you scan it in five seconds before it goes live. Most people start with a review step for the first week, then switch to fully automatic once they trust the quality. You stay in control of exactly how much you want to review.
Can it handle video as well as photos?
Yes. The vision model reads video frames, not just still images. It samples a few key moments in the clip, identifies the main subject and the tone, and produces a description that the caption writer can use. Video posts tend to produce slightly longer descriptions because there is more to see, but the caption output length stays the same.
Does each account need its own setup?
You set a goal, a tone, and a call-to-action destination once per account. After that the machine handles everything automatically for that account. If you have twelve accounts pointing to three different products, you set up twelve goal configurations and then leave them alone. The accounts run in parallel and the machine never mixes up which caption goes where.
Do I need to code to build one?
No. The whole chain connects a vision model, a caption writer, and your social scheduler using no-code tools. If you can connect two blocks together, you can build it. That is exactly what Automations Made Easy teaches.
More Money Makers like this one
Built from the same handful of mechanics. Each one captures, converts, or earns money that would have walked out the door.
Build the caption machine yourself, or work with me directly.
If you want to learn the mechanics and build it on your own stack, Automations Made Easy is the playbook. If you’d rather work with me one-on-one to plan it for your specific business, I take a small number of consulting clients each month.
€497 one-time · Lifetime access · 1,000+ students
