The 30-minute tool that makes better prompts for ads, posts, and landing pages

I’ve been using AI for years now.

I’ve integrated it into most of my projects. I’ve used it for content, ads, automations, internal tools, research, scripts, landing pages, image prompts, and most of the repetitive work that used to slow my businesses down.

And I still hate it with all my heart.

Not because it’s useless. It is obviously useful. The problem is that I’m a software developer at heart, and after 17+ years since writing my first line of code, my brain is trained to expect precision.

When you write code properly, you expect the same input to produce the same output. If you write:

1 + 1

You don’t expect the computer to say:

“Probably 2, but here are five emotionally compelling alternatives.”

Code is deterministic. Same input. Same output. No unexpected mistakes.

AI does not work like that.

You can give the same prompt to the same model three times and get three different answers. Sometimes one is useful, but most of the time you get generic lifeless outputs.

This becomes especially painful when you use AI for marketing: Facebook posts, image prompts, video scripts, landing pages, ad copy, email sequences, website sections, lead magnets, offer explanations, and all the other pieces of marketing where you need creativity, but you also need accuracy.

Here’s where most people go wrong.

They open ChatGPT and write something like:

Write me a Facebook post about my free course.

Or:

Create an image prompt for this quote.

Or:

Write a landing page for my product.

Then they’re disappointed when the result sounds generic, but the AI did its best based on what it received.

AI needs way more context than you think it should need.

Way more.

You expect it to infer things that feel obvious to you. You assume it will understand that a Facebook post for your personal profile should sound different from a paid ad. You assume it will know that a landing page for a free course before a booked-call funnel needs a different angle than a landing page selling a $7 audit. You assume it will understand the business model, the audience temperature, the emotional state of the reader, the offer mechanism, the brand voice, and the real purpose of the creative.

Sometimes it technically can figure out parts of this on its own.

But it often won’t.

Not reliably. Not consistently. Not in the exact way your business needs it to.

So it fills in the blanks with the safest possible average answer.

And this is where bad AI marketing begins.

A post meant to build trust is not the same as a post meant to generate booked calls.

Same topic.

Completely different context.

Completely different output.

Most people ask AI to create before they make it understand.

That’s why the first draft usually sounds flat. Not because the model is weak, but because the prompt is weak.

Think about how a good creative director works. They don’t hear “make an ad” and instantly start designing. They ask questions first.

Who is this for? What do they already believe? What are they afraid of? What are they tired of hearing? What is the offer? Where will this be used? What action should it create? What tone should it avoid? What examples are close to what we want? What would make this feel off-brand?

AI needs the same thing.

And to be fair, most serious entrepreneurs don’t prompt that badly anymore. They don’t just type one vague sentence and expect a premium creative to fall out of the machine.

But even when the prompt is decent, AI still needs more context than you think. Much more.

This is even more true now that the latest models can handle huge context windows. If the model can read hundreds of pages, or even close to a million tokens of context, there is no reason to starve it with a thin brief and then complain that the output feels generic.

Give it the offer. Give it the audience. Give it the funnel. Give it the positioning. Give it what to avoid. Give it the messy details you think are obvious.

Because obvious to you does not mean obvious to the model.

The shiny object version of this is the usual internet advice:

“Here are 500 ChatGPT prompts for marketers.”

There is another piece of advice you see all over the internet:

“Create a Google Doc with your best prompts and reuse them.”

That is not wrong. In fact, it can work very well.

But only if you understand why a prompt worked.

A prompt that worked beautifully for a landing page might fail completely when you use it for a YouTube video script, because those two tasks need different context.

For a landing page, the AI needs to understand the offer, the promise, the CTA, the objections, the proof, the funnel stage, and what the visitor needs to believe before taking action.

For a YouTube script, it needs a different brief: the hook, the pacing, the viewer’s current belief, the retention pattern, the story arc, the examples, and what should happen in the first 30 seconds.

Same business.

Different creative job.

Different context.

So yes, build a prompt library. But don’t treat prompts like magic spells. Treat them like briefs for specific types of work.

And this is where AI becomes painful in a very practical way.

Let’s say you have a good idea for a post while you’re between calls, or you need to write a few emails for a sequence before the day explodes into meetings, team questions, client messages, and the other thousand small things that come with running a business.

The idea is there. You know there is something useful in it. Maybe it’s a story from a sales call, a pattern you noticed in the market, a customer objection, a new angle for an ad, or a better way to explain your offer.

But now you have to sit in front of the laptop and translate that idea into a full AI brief.

Who is this for? What stage of awareness are they in? What do they already believe? What is the funnel context? What should the tone be? What should it avoid? What examples should it follow? What is the CTA? What is the business model behind it? What is the real point of the piece?

And you’re supposed to remember all of that while also doing actual work.

So you write the best prompt you can. You add some context. You explain the idea. You include a few constraints. Then you run it, wait for the output, read through it, and realize the result is not completely wrong, but it’s not completely right either.

Which is almost worse.

Because now you don’t have a clean creative. You have a draft that needs fixing. The angle is close, but not sharp enough. The tone is slightly off. The CTA doesn’t match the funnel. The example is generic. The image prompt misses the visual style. The email says the right thing, but not in a way that sounds like you.

So the task you wanted AI to make easier has turned into another editing job.

That is exactly the problem I wanted to solve with the small internal tool I’ll explain in a second.

The idea is simple: instead of starting with a blank prompt box, you start with a goal. Then the tool acts like a creative strategist and asks you specific, direct questions until it has enough context to write the actual prompt.

That matters because it is much easier to answer a good question than it is to mentally reconstruct every angle of the brief from scratch.

A good question pulls the context out of you.

A blank prompt box asks you to become the strategist, copywriter, creative director, funnel architect, and brand police at the same time.

That’s too much friction.

Especially when you’re busy.

So the tool reduces that friction before the creative work even begins. It does not try to magically write the perfect post, email, landing page, or image prompt from a half-formed thought. First, it helps you extract the missing context.

It asks one clear question at a time. You answer. It asks the next one. You answer again. Eventually, the brief becomes complete enough for the AI to produce something useful.

Not perfect.

Useful.

And that is the point.

Because once the context is clear, the next step is obvious: turn that context into a much stronger prompt.

So I built a small internal tool.

Nothing fancy. I vibe-coded it in about 15–30 minutes, and the app does one simple thing: I give it the goal of the prompt, then it asks me questions until it decides it has enough context. Only after that does it write the prompt for me.

That’s it.

But that small change makes a massive difference.

Instead of staring at a blank screen and trying to remember every useful detail, the system creates the context for me. The app acts like a creative strategist before it acts like a prompt generator.

For example, if I write:

I want to generate Facebook post ideas for my programming education brand.

It might ask who the audience is, whether they are complete beginners or people who already tried learning, what fear we are addressing, whether the goal is engagement, trust, comments, leads, or sales, and whether the post should sound personal, educational, controversial, funny, or motivational.

It might also ask what beliefs we want to challenge, what the post should avoid saying, how long it should be, and whether it should include a CTA.

Now compare that with:

Write me a Facebook post about learning programming.

Those are not the same task.

One gives AI a job.

The other gives AI a direction.

Big difference.

And this works for almost any kind of creative goal: Facebook posts, image prompts, video scripts, landing pages, ad copy, emails, sales pages, or a rough idea you need to turn into something useful before it disappears from your brain.

The format stays the same:

goal -> questions -> context -> better prompt -> better output

Funny side note: this pattern would probably improve meetings, sales calls, hiring, and relationship arguments too. But asking three good questions before producing a confident answer might still be too advanced for humanity, so we’ll start with prompts.

Here’s the simple architecture.

Instead of:

idea -> prompt -> output -> disappointment

Build:

goal -> questions -> context -> prompt -> output -> review -> reusable history

That’s the system. It’s small, but it matters.

You can build this yourself in 15–30 minutes with your preferred vibe-coding tool: Claude Code, Codex, Antigravity, Cursor, or whatever you’re comfortable with.

The instruction can be simple:

Build me a small app where I first enter the goal I want to accomplish. The app should have a context-gathering agent that asks me questions until it has enough detailed information. If one answer is incomplete, the agent should ask follow-up questions before moving on. Once it has the full context, it should pass that information into a separate prompt that generates the final result. Add chat history so I can reuse old context later, and make the final prompt or final output easy to copy.

The first version probably won’t be perfect. That’s fine. With these tools, you don’t just “order” software once and accept whatever comes out. You dialogue with the AI developer. You test the app, see where it’s imprecise, then ask it to adjust the questions, the logic, or the final prompt until it works the way you want.

And if you don’t know how to run it, literally write that into Claude Code, Codex, or whatever tool you use:

I don’t know how to run this app. Give me step-by-step instructions for a non-technical person.

That’s usually enough to get moving.

After you use this for your typical creative tasks a few times, you end up with something much more useful than a random list of “500 AI prompts.”

You end up with your own prompt library.

Prompts for your Facebook posts. Prompts for image generation. Prompts for email sequences. Prompts for landing pages. Prompts for video scripts. Prompts that already contain the context, rules, examples, tone, positioning, and constraints that matter for your business.

Then, the next time you have an idea, you don’t start from zero.

You copy the right prompt, paste it into ChatGPT, Claude, or whatever model you use, and add the fresh idea at the bottom.

Something like:

Here is the new idea I want to turn into a post: ...

That’s where the leverage is.

The machine writes the heavy prompt for you once. You test it, improve it, and save it. Then you keep reusing it instead of rebuilding the full brief every time you want to create something.

And because the prompt already carries the important context, the output is usually much closer to what you wanted from the first try.

The ROI of a tool like this is not just that it saves 30 minutes of editing every time you use AI.

The real ROI is consistency.

You get better first drafts, reduce the number of unusable outputs, preserve brand context, make creative production easier to delegate, and stop relying on the founder to personally explain the business from scratch every single time.

That last one is important, because a lot of businesses don’t have a content problem. They have a founder-context-transfer problem.

The founder has the positioning in their head. The AI doesn’t. The team doesn’t fully have it either. So every creative task becomes a guessing game.

A context-first prompting tool fixes part of that.

Not all of it.

But enough to matter.

That’s the whole point.

AI works better when you stop treating it like a magic box and start treating it like a system.

For creative work, that system can be simple: ask better questions, collect better context, reuse better prompts.

If you want help building this kind of practical AI workflow for your business, go to Saleswell.us.

And if you know someone who needs this, send them our way. We have a referral program too.