Stop Wrestling With AI Prompts. Start Training an Assistant That Understands You
The problem with prompts, and how to fix it forever
We’ve all been there.
You load up your favorite AI writing tool, paste in a carefully crafted prompt you found on some thread, and hit generate. The output comes back—decent, maybe even passable.
But you read it and think: "That’s not quite what I really wanted."
So you tweak the prompt. Add more context. Specify tone. Provide examples. Try again.
It’s better. But still, it doesn’t feel fully like you.
And there you are—stuck in this strange game of prompt engineering whack-a-mole, endlessly adjusting the words you feed into the system hoping for magic on the other side.
This is the fragile relationship many creators have with AI right now: prompt, tweak, retry, repeat. But what if I told you there’s a better way? One that doesn't leave you trapped endlessly wrestling with prompts?
Author’s Note:
This edition has a sponsor! Turn your writing into income 5x faster with WriteStack.io. More about it below.
The fragility of prompting
At first glance, prompting feels empowering. You feed the AI a few well-chosen words, and it generates something impressively coherent.
But with each new project, you quickly realize you're starting from scratch every time. The AI doesn't know your voice, your frameworks, or your deeper purpose—forcing you to manually inject context with every prompt.
As you try to improve, you fall into the exhausting cycle of prompt perfectionism.
You invest hours refining intricate instructions, but the outputs still feel generic, disconnected from your authentic voice, and often require heavy rewriting.
And as AI models improve, more people master the same prompt tricks, making it harder to stand out. Relying on prompt crafting alone doesn’t make you future-proof; it makes you increasingly fragile.
Shift from prompting to training your AI assistant
It’s time to stop seeing AI as a vending machine that spits out words when you punch in the right combination. Instead, treat it like a long-term creative partner you can train.
The real power is in building an AI assistant that knows you. One that understands your voice, your frameworks, your audience, your weird quirks, and your deeper mission.
This isn’t about stacking one clever prompt on top of another. It’s about building context—and building it continuously.
1. Create your four foundations
Before your AI assistant can create anything meaningful, you need to give it solid ground to build on. These four foundations are the core of that process:
Personal Profile. Define your core values, your worldview, and your key stances. What do you believe? What are the non-negotiables that shape your voice? This is what gives your writing its moral spine and philosophical clarity.
Business Case. Establish your strategic positioning. Who are you writing for? What niche are you serving? What unique angle or promise does your brand stand on? This tells AI who you’re speaking to and why your voice matters in that space.
Writing Profile. Document how you write. Capture your sentence rhythms, your tone, your favored structures, and the little quirks that make your voice recognizable. Define your writing principles: how you approach argumentation, how you simplify complex ideas, and how you emotionally connect with your audience.
Writing System. Build your repeatable workflows. Gather your preferred templates, outlines, formatting preferences, and ideation frameworks. This allows your assistant to mimic how you build content from idea to finished product, not just the end result.
If you want to learn the specific frameworks, exercises, and prompts for building your Four Foundations, check out my ebook, Authentic AI: How to Create a Hyper-Personalized Content Assistant that Thinks and Writes Like You.
It gives you the full system to create your hyper-personalized assistant step-by-step.
2. Feed it your body of work
With your foundations set, it's time to let your AI study your real-world writing. Feed it your full body of work: published articles, newsletters, social posts, client deliverables, even personal emails or DMs that reflect your natural communication style.
You're not just feeding it topics or keywords—you're letting it absorb how you think and express ideas. This teaches your AI assistant your narrative style, your pacing, your favorite phrases, and even how you transition between ideas. The more material you feed it, the richer its understanding of your voice becomes.
3. Create iterative feedback loops
Even with strong foundations and plenty of data, your AI won’t nail your voice perfectly right away. That’s why continuous feedback is critical.
Engage with your AI like you would with a junior writer. Review its drafts. When something feels off, don’t just fix it—explain why it’s off. Highlight where it missed your tone, nuance, or pacing. Break down what made certain word choices feel wrong or why a transition didn’t fit your usual rhythm.
Every correction becomes a new training data point. Over time, your assistant sharpens its understanding and aligns closer to your authentic voice. Use the SRR framework as your guide:
Spot the issue, and clearly articulate it
Refine the output by giving your AI assistant specific points of improvement
Reinforce the lesson by encoding persistent fixes into your document
The more you collaborate, the more your assistant evolves into a true extension of your creative process.
Does prompt engineering have a place?
Yes, it absolutely does—but prompt engineering is tactics, not strategy.
Once you’ve trained your AI assistant to deeply understand your voice, values, and writing system, that’s when prompt engineering becomes powerful. Because now, you're not prompting into a cold, clueless system. You're giving targeted instructions to an assistant that already knows how you think.
At this stage, crafting, refining, and saving specific prompts allows you to amplify very focused tasks: generating headlines, building outlines in your preferred structure, etc.
The difference is huge.
Instead of wrestling with prompts just to get AI to sound passably like you, you’re now leveraging prompts to magnify your creative intent—10x-ing both the speed and the alignment of your outputs.
You create a compounding creative asset
When you shift from wrestling with prompts to training your assistant, everything changes.
You eliminate the constant starting-from-zero feeling. You stop wasting hours rewriting bland drafts. You move into a workflow where you and your AI collaborate seamlessly, generating first drafts that feel 80% aligned from the start.
And most importantly, your AI becomes more valuable over time.
The more you train it, the better it understands you. Every project adds new layers of depth. What starts as an assistant becomes a compounding creative system tailored to your voice and mission.
But …
"I don’t have time to train an AI assistant."
You’re already spending time rewriting and fine-tuning drafts. Training simply moves that time investment to the front of the process, where it pays dividends long-term.
"Isn’t this too complicated?"
No more complicated than spending hours crafting prompts that still require rewriting. In fact, once trained, your assistant simplifies your workflow significantly.
"Will AI updates break my trained assistant?"
While models will evolve, your stored context, voice patterns, and unique frameworks remain durable assets you can reapply to newer systems.
The new writing game
Prompt hacking belongs to the old game.
The new game is about creative leverage: building a personal AI system that knows you intimately and grows with you over time.
You’re not just creating content faster. You’re building something antifragile—a system that compounds your thinking, preserves your voice, and allows you to focus on what matters most: bringing your most original ideas to life.
A word from our sponsor:
Most Substack writers ignore Notes.
Big mistake.
Top creators get 50% of subs from them.
WriteStack gives you personalized Notes + a dead-simple scheduler so you can post consistently, grow faster, and never run out of ideas.
And an all-in-one notes dashboard, so you can double down on what works, and ditch the others.
Really enjoyed your approach to moving beyond prompt wrestling; the Four Foundations framework is much more sophisticated than the usual "better prompting" advice floating around.
I'm curious about the longer term viability of this model, especially with infrastructure costs scaling up so dramatically. When you're advising people to upload their entire corpus and build these deep feedback loops, how are you thinking about the compute economics behind personalized AI assistants?
Jensen Huang from Nvidia mentioned we'll need 100x more compute power for next gen reasoning models. Given that data centers already account for 44% of new electricity demand growth, I'm wondering when you think truly personal AI assistants will be available that aren't just API access to cloud LLMs with rising usage costs.
Your training methodology is excellent for creating sophisticated AI relationships, but I'm trying to understand the business model sustainability when the underlying compute becomes way more expensive. Are you seeing this as a premium service tier, or do you think local inference will make personal assistants economically accessible for individual creators?
The irony is that the better we get at training these systems (using methods like yours), the more compute intensive and expensive they become to run. Curious how you're thinking about that trajectory for your subscribers.
Would love your take on the economics side.
When I read your introduction, I screamed, “that’s so me”!
I would like a copy of your ebook but not sure the link is directing me to the right place.