The Five Levels of AI Augmentation

How to know when you're using too much AI, or not enough

When I first started using AI for writing, I kept circling back to one question.

How much is too much?

I’d finish a piece, look at it, and wonder if I’d crossed some invisible line. Had I leaned too hard on the tool? Was this still mine? The question nagged at me, and I couldn’t find a satisfying answer anywhere online (for one, everyone’s wondering about it, too.)

So I started examining it myself. I wrote about it. I tested boundaries. This newsletter is one of the results of those inquiry, as my introspection became words and words became articles.

And in this article, I’ll share the answer I’ve arrived at myself.

The AI balance conversation nobody’s having

The loudest voices in AI fall into two camps.

On one side, the power users. They automate everything. They build agent swarms and content pipelines that run while they sleep. To them, resistance is a crumbling sandcastle of nostalgia.

On the other side, the skeptics. They refuse to touch AI at all. They see it as theft, as creative betrayal, as the death of something sacred about making things by hand.

Both sides have their points.

But I think, neither has the full picture.

Most creators I know don’t live at either extreme. They’re somewhere in the middle, using AI for some things, avoiding it for others, depending on the situation. What we all have in common, though, is that we’re constantly second-guessing whether we have the right AI content creation balance.

The advice they usually get is frustratingly vague. “Use AI just enough.” “Find your comfort zone.” “Don’t lose your voice.”

That’s not helpful.

What’s missing is a map. Something that shows the terrain between “no AI at all” and “AI does everything.” Something that helps you figure out not just where you are, but where you actually want to be.

A framework for finding how much AI to use

I wrote about this topic last year, proposing three levels of AI augmentation. The article did well on Medium. But as I’ve worked more deeply with AI since then, three levels started feeling too crude. The landscape I’m in right now is more nuanced.

So here’s the refined version. Five levels of AI augmentation.

As you read through the AI augmentation levels, think about your own process. Consider the different stages of your work, whether that’s content creation, research, client delivery, or something else entirely. Each stage can sit at a different level. You don’t have to pick one and apply it everywhere.

The goal isn’t to reach the highest level. It’s to know which level is right for which part of your work.

First level: Unassisted (zero AI involvement)

This is zero AI involvement. No research assistance, no drafting, no editing suggestions, no image generation. Nothing.

Some things belong here.

Think about the content that’s most intimate to you. Personal letters. Journal entries. Video messages to people you love. The kind of creation where the entire point is that you made it, every word, every choice, every imperfection.

For some creators, this level extends across their entire process. Everything they produce is a full expression of themselves. Any AI involvement, no matter how small, feels like contamination. I’ve met writers like this. They’re not being dramatic. For them, the process is the product.

I think of this as your “core self” in creation. The parts that are inviolable. The parts where using AI would defeat the purpose.

Your core can be large or small.

A large core means most or all of your creative process falls into this category. You want human hands on everything.

A small core means only a sliver of your process is sacred. The rest is fair game for assistance.

I have a small core. For me, it’s ideation. I want the concepts I develop, the frameworks I name, the problems I choose to explore, to come from my own thinking. That’s where my identity lives. The rest of the process, I’m happy to augment.

Knowing the size of your core matters. It explains why some creators embrace AI readily while others recoil. Neither response is wrong. They’re just working with different-sized cores.

The question to ask yourself: What parts of your creative process are you, and you alone? What would feel like a betrayal to outsource?

Second level: Augmented (light AI assistance)

This is light assistance. You’re using AI for small, specific tasks that reduce friction without changing how you think or what you produce.

At this level, AI might:

• Summarize a long article so you can decide if it’s worth reading

• Generate a few headline options when you’re stuck

• Clean up rough notes into something readable

• Create a simple image for a presentation

• Answer a quick factual question

The keyword is “assist.” You’re still doing a lot of the thinking. AI is handling the grunt work, and at the very edges, at that. More or less, you’re treating AI like another version of Google. Probably one that can make memes for you.

Regardless, this level still matters because not everything needs to be handcrafted or fully automated.

There’s a middling ground where small conveniences add up without fundamentally changing your output. For beginners, this level builds confidence. You start to see what AI can handle without it taking over.

But there’s a subtle risk, too.

Once you get used to the convenience, you might lose patience with doing anything manually. The threshold for “this is tedious” drops lower and lower, and that can get very dangerous, very fast, as you “meddle with powers you don’t understand.”

There are real rewards, though. Less cognitive load. Faster execution on the small stuff. That boring email? Yep. Boring news article? Can’t really be bothered to look up what kind of bird it was that appeared in that specific movie (but you’re really curious, for some reason)? Yep, and yep.

The question to ask yourself: What parts of your process is AI lightly touching? Are those the right parts, or have you drifted into convenience without noticing?

Third level: Accelerated (moderate AI involvement)

Here, AI becomes a significant part of your workflow. It’s not just handling edges anymore. It’s doing substantial work.

At this level, you might be:

• Having AI research and draft 70% of a newsletter

• Generating full customer personas from data

• Building complete outlines for major projects

The key difference from the second level is depth of integration. AI isn’t assisting anymore. It’s executing.

At this point, you’ll probably have paid AI tools for creators, custom prompts, integrated AI writing workflows, and a pretty good understanding of what AI can and can’t do for you.

This is where the productivity gains get serious. What used to take a full day might take two hours. What required a team can now be done solo.

But the risks are proportionally larger.

In marketing, your voice can flatten. When AI handles most of the writing, your distinctive patterns start to smooth out. You sound competent but generic.

In research, your thinking can weaken. When AI does the research and structuring, you stop wrestling with ideas yourself. The mental muscle atrophies.

Your creativity can suffer. The struggle that produces original insight gets bypassed. You end up with efficient content that lacks depth.

These aren’t inevitable outcomes. But they’re very common ones.

The reward, when done carefully, is capacity. You can do more, serve more people, build more assets. For scaling teams or delivering consistent output, this level can be transformative.

The question to ask yourself: What are you gaining at this level? What might you be losing? Have you checked recently?

Fourth level: Reflected (personalized AI systems)

This is where things get interesting.

At this level, your AI systems are personalized to your identity. There’s human-AI collaboration. They know your voice, your values, your mission. They’re trained on your past work, your thinking patterns, your frameworks.

Instead of generic output, you get output that sounds like you. Instead of AI replacing your perspective, it mirrors it.

This is the Authentic AI approach I’ve been developing. You build profiles that capture your personal philosophy, your writing style, your business goals. The AI draws from these to produce work that aligns with who you are.

Other writers and thinkers have different formats. But that’s the gist of it -- your AI reflects a lot more of who you are.

At this level, you might have:

• Custom GPTs trained on your published writing

• Content guidelines embedded in every AI interaction

• Systems that reference your mission before generating anything

The result is unpromptable work at scale. Your output remains distinctive even with high AI involvement.

The setup requires discipline. For writing content, for example: You have to articulate things you might never have made explicit before. What do you actually believe? What makes your voice yours? What’s your mission, specifically?

The risk is overconfidence. Once you’ve built these systems, you might assume AI knows you perfectly. It doesn’t. It has an approximation, a ghost even. The approximation needs regular updates as you evolve, and AI is going to get it wrong a lot of times, too.

The reward is alignment. Your content, your assets, your systems all pull in the same direction. You maintain creative clarity even as you scale.

The question to ask yourself: What AI-related processes do you have that are customized to your specific beliefs and ideas? If the answer is none, why not?

Fifth level: Unpromptable (full human-AI collaboration)

This is full integration. AI becomes an extension of your creative and operational self.

At this level, you’re not just using personalized AI systems. You’re building interconnected systems. Your voice, mission, business case, and knowledge base all feed into a second brain that AI draws from to create new things.

You’re using:

• Multi-tool pipelines that repurpose and distribute content automatically

• Agents with full knowledge of your concepts that can interact with leads

• Assessment tools that take user input and provide custom recommendations based on your frameworks

• Automation systems that pull content ideas, filter them through your concepts, and creates the content -- all while you’re focused elsewhere

I built a system like this for my own work. One Substack article becomes eleven pieces of content across platforms. The system knows my voice, my topics, my audience. It produces material I’d be comfortable publishing with minimal editing.

The practical application question I ask clients is what I call the Helpful Alien Question.

Imagine an advanced civilization shows up tomorrow, willing to help with anything. They’re superintelligent, infinitely capable, and all they need is for you to tell them exactly what to do.

What problems would you have them solve in your business?

Would they handle your marketing videos? Your SEO? Your customer support? Your content repurposing?

Your answers reveal where you’re ready for this level of augmentation. And where you’re not.

The risk is over-engineering. You can build systems so complex that maintaining them becomes a job in itself. You can also lose touch with your core if your boundaries aren’t clear. Identity drift happens when these systems evolve without intentional oversight.

The reward is maximum leverage. Scalable influence. The ability to build signature assets at high speed while staying aligned with your mission.

The question to ask yourself: If you could automate anything with perfect fidelity to your vision, what would it be? And are you ready for that?

Finding your own balance

These five levels aren’t a ladder to climb. They’re a map to navigate.

Some parts of your work should stay at level one forever. Some parts might benefit from jumping straight to level five. Most will land somewhere in between.

The point is to make conscious choices about where each part of your process belongs.

Here’s how to use this framework:

First, list the stages of your work. Research, ideation, drafting, editing, publishing, distribution, whatever applies to you.

Second, identify where each stage currently sits on the five levels. Be honest about this. Where are you actually using AI, and how deeply? Then, ask whether that’s where it should be. Is your core protected? Are you getting the benefits you want? Are you aware of the risks you’re taking?

Third, adjust deliberately. Move stages up or down the levels based on what you want to preserve and what you want to scale. And ensure you can verbally articlulate the why.

The goal isn’t maximum AI involvement. It’s maximum alignment between your tools and your values.

Some creators will read this and realize they’re using too much AI for their comfort. Others will realize they’re leaving leverage on the table. Both are useful realizations.

The question was never “how much AI is acceptable.” The question is “how much AI is right for you, in this part of your work, given what you’re trying to build.”

Now you have a map to answer it.

---

If you want to build personalized AI systems that reflect your voice and mission, that’s what I help founders do through the Pathfinder and Builder programs. Reach out if you’re ready to stop guessing and start building with intention.