How I Get Professional Results With Any AI Tool (Even New Ones)
How to build a repeatable system for working with any AI tool and get professional results - even if you're just starting out.
Happy New Year, Boomies! 🎉 Hope that this year brings you everything you’re building towards.
I kept thinking about this thing I do that I’ve never fully explained and it felt like the right first post for this year.
Because if you're starting 2026 with intentions to get more serious about AI or you're already using it but want to level up, this is the approach that will get you to actually good results with (almost) any AI tool, faster than anything else.
(Speaking of intentions - if you missed my post on how I think about resolutions, you shouldn’t skip it. The prompt I shared works both if you’ve already set them or you’re just doing it now, and it will help you challenge those goals until you're sure they'll actually stick this year.)
Let me show you what I mean.
The problem with learning how to work with new AI tools
Let’s say you want to start using Midjourney. Or you’re already using it but your results are... meh.
What do you do?
You probably search YouTube. Read a few articles. Maybe scroll through X for tips. You’re piecing together advice from different people, and that takes time and a lot of experimentation. It makes sense to do it this way. All of these things help, because they give you earned experiences from others one step ahead of you.
But sometimes, you want to get good results much faster. Without skipping the learning.
That's what I want to show you today.
Most tutorials give you the dish, I want to give you the recipe:
The system I now use to get good results with everything - learning Midjourney, prompting Nano Banana Pro, getting better at vibe-coding tools like Lovable, working with different AI models.
Once you see it, you can’t unsee it. And you can use it to create your own “recipes” for working with any AI tool you’ll ever need to learn.
It's the process behind how some of us in the AI space actually work, but rarely show. Master this, and you'll know how to build your own shortcuts for any AI tool you touch.
If you've been thinking about becoming a premium subscriber, the 20% discount ends in one week. It's the lowest price this publication will ever be.
The core idea behind my simple system
Instead of just consuming random tutorials, you build yourself a system for working with these tools, using official documentation + AI.
You’re essentially creating a personal expert that knows all the best practices, so you can ask it questions whenever you need, instead of searching every time.
I’ve done this multiple times:
When I built The Lovable Prompt Architect - a CustomGPT trained on the official Lovable playbook that generates prompts structured exactly how Lovable needs them to perform best (I documented the whole process on the LAB)
When I created the master prompt for Nano Banana Pro that generates image and infographic prompts formatted the way the tool actually expects them
This is the method that accelerates how fast I get good results. But it also helps me actually learn the tools, because I see the prompts, repeatedly create them, and start to understand their structure and the way they should look.
Here’s the exact process.
Step 1: Find the official documentation
This is the unsexy step everyone skips. But it’s the most important one.
Who can teach you better about a tool than the people who built it?
Official documentation exists for almost everything, it’s just that most people don’t think to look for it or they find it too dry to read (it really is, I know).
But that’s fine. You don’t need to read it. You need to collect it.
Here’s where to find official docs for the most common AI tools:
Prompting LLMs & Using different AI models
OpenAI Academy – How to prompt ChatGPT
OpenAI Cookbook - Working with the OpenAI API
Google’s Prompting Guide – Official guide for Gemini
Google’s Prompt Design Guide - Working with Gemini API
Anthropic Academy – How to work with Claude
Claude Docs - Working with the Anthropic API
Creating Images with AI
Midjourney Getting Started – How to prompt Midjourney
Google’s tips for using Nano Banana - Prompting Nano Banana Pro
Image generation with Gemini API - Working with Gemini API for creating images with Nano Banana (Pro)
Nano Banana Pro Examples – Over one hundred of image prompt examples
ChatGPT Images - Prompting guide
Creating Videos with AI
Veo Prompting Guide – Google’s video model
Sora 2 Prompting Guide – OpenAI’s video model
Vibe-coding with AI
Lovable.dev: Prompting Guidelines, Prompting Playbook, Debugging
Claude Code: Getting started
I compiled a much more comprehensive list of free AI learning resources in my recent guest post for Michael Spencer’s AI Supremacy :
If you’re just beginning your AI journey in 2026, that’s a great place to start.
Worth noting: these resources aren’t always called “courses”. Sometimes they’re PDFs, sometimes they’re help docs, sometimes they’re scattered across multiple web pages. That’s what makes them hard to follow, but also why this method works so well. And you're not limited to official documentation. You can add any guides, tutorials, or courses you find useful.
Step 2: Consolidate everything into one place
Now you need to save all that documentation somewhere you can use it.
There are a few ways to do this:
Option A: Copy-paste into a document
If the documentation is on a webpage, just copy the entire page into a Word doc or Google Doc. Yes, really. That’s it.
You can combine multiple sources this way, just add some headers to keep things organized.
Here’s what mine looked like for Lovable, which englobed the official Prompting Guidelines, the Prompting Playbook, the debugging flow.
Option B: Use existing PDFs
Sometimes you get lucky and the documentation already exists as a PDF. Like Google’s Gemini Prompting Guide. Even less work for you.
But of course, you can gather multiple sources, that combine a lot more documentation from all official AI models so you have a complete view of how to work with any AI model.
Option C: Use NotebookLM for multiple sources
If you have more than 2-5 sources, drop them all into NotebookLM and, optionally, create a synthesis using the reports feature (which I deep-dived on in this article):
“Create a comprehensive report (around 5000 words) extracting the key principles, best practices, and practical guidelines from all sources.”
Now, this is a generic prompt. Yours should be more specific depending on what’s inside your sources and what you want to extract.
If you're collecting prompting guidelines from multiple AI models because you want to build a system that writes prompts for you, for example, ask explicitly for the report to focus on extracting the prompting principles (what makes a great prompt, what structure works best, what to avoid).
By the end, you’ll have one document with the distilled wisdom of everything.
This is where the real magic happens. In the next section, I’ll show you how to turn this documentation into a reusable system - whether that’s a CustomGPT, a Gem, a NotebookLM project, or just a really good prompt.
I’ll walk through the exact process I used, including the actual instructions and workflow that make it work.
P.S. If you’re not yet a paid subscriber, this is a good week to join - there’s still a 20% discount available. You’ll get access to the full LAB with all my documented processes, plus every deep-dive like this one.
Step 3: Build your learning system
You have the documentation. Now you need to make it usable.
There are three approaches depending on how often you’ll need this: