The 7 Levels of AI User (and how to level up)

You can have two people using AI every single day. One of them is saving a few

single day. One of them is saving a few minutes on emails. The other is running entire business systems that operate while they sleep. The difference isn't intelligence or technical skill. It's

knowing which level they're at and what to do next. Thanks to HubSpot for sponsoring this video. This video will show you all seven levels where you fit

and exactly what to do next. Moving up

happens faster than most people expect.

This is where everyone starts. You have

a free account in chat GPT and you're basically using it like a fancier search engine, asking it questions you used to Google, maybe getting it to write a quick email. It's useful, but it doesn't

quick email. It's useful, but it doesn't feel like a superpower yet. You're here

if you mostly ask one-off questions and don't really think about how you're asking them. The shift to level two is

asking them. The shift to level two is simple. You have to start noticing that

simple. You have to start noticing that how you ask changes what you get.

Now you've had the moment of realization where you asked something two different ways and got wildly different results.

Now you're actually thinking about your prompts. A couple big unlocks here are

prompts. A couple big unlocks here are giving context and including examples. A

simple prompt structure is instruction, context, and constraints. Then when

applicable, give it an example of an output you liked or a result you're looking for. Now that's good to know,

looking for. Now that's good to know, but there's two shortcuts to just speedrun this whole level. First is

asking the AI to ask you questions before it answers to gather needed context or clarify the intent. It knows

what it needs better than you do and it will ask you a list of questions to gather all of it. The response you get back will be exponentially better than what you would get otherwise. The second

way is you'll often have a long back and forth before you get what you want. Just

kind of realizing along the way that you didn't specify something or needed additional context. Once you've done

additional context. Once you've done that back and forth, ask it to write the prompt that would have gotten you there on the first try. If it's a task you'll be repeating, save that prompt so you can skip all that back and forth time in

the future. Those two tips will give you

the future. Those two tips will give you dramatically better results, basically right away for very little effort. To

get to level two, use the basic prompt structure instruction context constraints. Ask it to ask you questions

constraints. Ask it to ask you questions before it answers. And when an answer misses the mark, don't just accept it.

Rewrite the prompt with more context and try again. That habit alone will get you

try again. That habit alone will get you here fast. Once you're regularly getting

here fast. Once you're regularly getting useful output and feeling the value, you'll hit the next wall, starting fresh every single conversation and rebuilding context from scratch every time.

Breaking through that is pretty simple.

This is where you see a real noticeable change in your productivity. Not just

getting better answers, but tasks that used to take you an afternoon starting to take 20 minutes. You stop getting stuck on things and it feels intuitively woven into your workflows. This involves

going deeper and utilizing all the features available. This helps in many

features available. This helps in many ways, but one big unlock is with context engineering. Providing the right context

engineering. Providing the right context gets you much better results, but you don't want to have to start from scratch in every new chat, feeding in all that context. Projects solve this problem.

context. Projects solve this problem.

Chat GPT or Claude let you build dedicated workspaces. One for work, one

dedicated workspaces. One for work, one for creative stuff, finances, health, whatever you need. each with their own instructions, tone, context files, and formatting preferences already baked in.

You stop reexplaining yourself every chat. Your work project already knows

chat. Your work project already knows your brand voice and SOPs. Your writing

project already knows your style. Your

fitness project already knows your goals. You just open a new chat and go.

goals. You just open a new chat and go.

This is also when you start testing the different tools like Claude, Gemini, and Chat PT have very different strengths.

And by now, you have enough experience to actually feel the difference. To get

here, pick your most common topic for AI use and build a project for it. Include

context files that are relevant and write out custom instructions. Anything

you find yourself reexplaining, make sure it's in there. You can definitely use chatpt to help write those. Just

kind of dump out all the things you think should be in there and have Chat GPT touch it up. Once you stop rewriting the same setup every time and have context baked in, you'll immediately feel the difference. You'll have a real

workflow with AI, but you're still operating almost entirely inside LLMs. We're about to get our first taste of AI agents, and as you can imagine, that's where a lot of this is headed towards.

I'll be going deeper into these as we progress through the levels. There's

been huge advancements over the past year, and is starting to actually live up to the hype, which is why I also worked with HubSpot to create a complete guide to AI agents updated for 2026.

This is a free resource that goes deep into agents with frameworks, worksheets, and implementation details. There's a

simple is this an agent job decision framework you can actually use to evaluate your workflows. A full

breakdown of a humans for judgment, agents for execution model with real examples, and an entire chapter of use cases across content and marketing, creator and entrepreneur workflows, and

business operations. plus a complete

business operations. plus a complete implementation roadmap with the step-by-step process for assessing what to automate. Integrating agents with

to automate. Integrating agents with your existing tools and data, adding the right guard rails and tracking ROI through efficiency, quality, and business impact metrics. We packed this with everything you would need to get

started and actually implement AI agents. It's free to download at the

agents. It's free to download at the link in the description. And thanks to HubSpot for partnering with me on this resource and for sponsoring this video.

Level four is when you realize although you've already gotten a lot of benefit, you're still just at the beginning. AI

is an entire ecosystem and different tools are built for different jobs. You

stop asking what can chatbt do for this and start asking what's the best tool for this. You're pulling in notebook LM

for this. You're pulling in notebook LM for research and working through large documents using granola to handle meeting notes automatically so you can actually stay present in conversations.

You've got image or video generation in your toolkit when you need it. Platforms

like Higsfield or Freepick let you access all those models in one place.

I'm not going to just sit and list off all the different tools and what they're good for. They go on and on, and I have

good for. They go on and on, and I have a lot of videos covering that, but one more I'll mention here. An easy way to get your first taste of a gentic AI is with Manis. Instead of you prompting

with Manis. Instead of you prompting back and forth, you describe a goal and it goes off and develops the plan for you and handles a multi-step process, orchestrating different types of models together all on its own. The first time

you see this is pretty mind-blowing. You

really take your hands off the wheel while it just works for you. And here in level four, there's also a deeper level to LLMs we haven't discussed yet. You

can start to build. Claude, Chachyp, and Gemini all have this feature. In Chachi

and Gemini, it's called canvas. In

Claude, it's called artifacts. You can

describe a tool you want. It will write the code and build it right there. Fully

interactive with zero experience required. You can prototype things in

required. You can prototype things in minutes. This could be an internal tool,

minutes. This could be an internal tool, a tracker, an interactive dashboard, or a website. and they can have a lot of

a website. and they can have a lot of functionality. This is just a taste of

functionality. This is just a taste of vibe coding, but when you try it, you see just how easy it is. From a few simple sentences of what you want, it can build it. Then you just ask simply for new features and changes. There's

tons of different tools you can experiment with. Each person will have

experiment with. Each person will have different tools that apply to their role. But the point here is everything I

role. But the point here is everything I mentioned is really easy to use. None of

these tools are technical whatsoever.

They're all unique and may take some experimentation, but it's fairly intuitive to just try something. Then if

it's not right, follow up with what you didn't like. It usually just fixes it.

didn't like. It usually just fixes it.

Then just over time, you get faster. But

now we're weaving multiple tools together to get things done in ways that weren't possible before. To get here, pick one tool that's not an LLM and actually use it for something real. I

have a lot of tool roundup style videos of the things I personally actually use.

This could be Notebook LM if you do research, granola if you're in meetings, Higsfield if you do creative work, and also open your artifact or canvas feature and describe a tool you wish

existed. Don't overthink it. The point

existed. Don't overthink it. The point

isn't to ship something polished. It's

to feel how fast you can go from an idea to a working prototype. When you're on this level, you're using several different types of tools regularly, but they're all still things you manually kick off and follow up, go back and

forth, and check on yourself. Level five

is where you start building things that run without you.

You've tasted what's possible and now you start to build intentionally.

There's a mindset shift here. You stop

seeing a tedious task and thinking, "How do I do this faster and start thinking, how do I build something that does this for me?" The platforms like Lovable or

for me?" The platforms like Lovable or Google AI Studio let you do a ton for free. These make it just as easy to

free. These make it just as easy to build internal tools as the canvas features you used in level four, but they are purpose-built for it. So, you

can create genuinely useful things.

Client dashboards, internal tools, custom workflows for your team. Real

stuff still without needing to be a developer or knowing how to read a single line of code. You can really start using your imagination to build out complex tools. And what's really cool is many of the tools you may have

been experimenting with can be built into the tools and apps you build for yourself. You can prompt for something

yourself. You can prompt for something that incorporates Cloud for writing and Nano Banana for generating images all in a unique interface and workflow purpose-built for you. At this level,

you also start automating not just one-off tasks, but full processes that get triggered on a schedule or by an event and run end to end without you touching them. Zapier is the easiest

touching them. Zapier is the easiest entry point here. They have a co-pilot that lets you just describe what you want an automation or agent to do and it builds the whole thing and we run it at Futureedia across our entire backend.

Everything that would otherwise take a full-time admin, keeping data synced across platforms, running background tasks, routing information between tools, Zapier just handles it. To get to

level five, start with one thing you do repeatedly and ask yourself what would have to be true for this to just happen automatically. Depending on what that

automatically. Depending on what that is, open Zapier and describe it to the co-pilot or open Google AI Studio and describe the tool. You don't need to know how to build it. You just need to know what you want. That first

automation that runs without you touching it is a big unlock. Once you

feel that, you'll start seeing opportunities everywhere. You're here if

opportunities everywhere. You're here if you've built at least one simple thing that runs on its own, but your more complex workflows still need you to prompt them or check in regularly. They

aren't streamlined and definitely aren't automated.

At this level, you don't see problems anymore. You see systems waiting to be

anymore. You see systems waiting to be built. Every repetitive process, every

built. Every repetitive process, every workflow that feels slow, every task that a human is doing manually, your brain immediately starts designing the solution. This is where claude code

solution. This is where claude code typically becomes one of your main building tools. You can build almost

building tools. You can build almost anything you can describe. I thought

this was a funny one I did for a previous video when everyone was talking about Cal AI being acquired. I took a couple screenshots from the App Store page and asked it to analyze those and build the app for me. I needed one

follow-up, then it was functioning perfectly where I could upload an image of food and it would identify all the ingredients and estimate the calories and macros and had long-term tracking

and everything. It's kind of crazy how

and everything. It's kind of crazy how easy it is to just replace software you use with a couple prompts. That's of

course using it for yourself. Managing

payments, scaling, or building communities is obviously harder, but that bespoke personalized software on demand is pretty cool, too. Now, I'm not going to walk through the whole terminal setup here. You can also just use it

setup here. You can also just use it right in the cloud desktop app with no additional setup. Just a couple of quick

additional setup. Just a couple of quick tips in cloud code is to use the planning mode when building from scratch. It will think deeply and

scratch. It will think deeply and develop the whole concept features and plan for everything first, then confirm it all with you. After that, just build feature by feature, testing each one

before moving on to the next. Then in no time, you'll have a fully functioning professional level piece of software you built yourself. This can be anything.

built yourself. This can be anything.

full AI powered web apps and mobile apps, internal tools and dashboards, Chrome extensions, bots, scraping tools, games, on and on. Another big tool here

is NADE. You can automate complex

is NADE. You can automate complex workflows that can even be agents that orchestrate other agents. Processes that

would normally take hours or require a team of people can run in the background continuously and without you involved.

We run these for some of our more complex workflows at Futureedia. And

then there's OpenClaw. everything up

until this point. You build it and it runs. Open Claw. You build it and then

runs. Open Claw. You build it and then it starts to think for itself. I want to be upfront about this one. It's really

interesting and there's a reason there's so much hype, but it is absolutely not for everyone. You're not even close.

for everyone. You're not even close.

Everything I've mentioned up until this point, anyone with a little motivation and mind for experimentation can learn.

I personally have zero coding background and use all of these. But Open Claw is another story. You can learn it if

another story. You can learn it if you're not technical, but it will not be as easy as the others. But if you don't know, this is an open-source personal AI agent that runs locally, usually on a

dedicated device like a Mac Mini, or you can use a VPS. You connect it to a language model, give it access to your tools, your email, your calendar, your browser. Some people even give it a

browser. Some people even give it a credit card, and it operates as a persistent, just always assistant that goes out and does things and just comes back and updates you when it's done. But

an important part is it can remember things across sessions. So, it's

learning context about you over time.

and improving. What makes it particularly unique is the interface.

Usually, this all happens from WhatsApp or Telegram messages. You have a multi- aent system, all with specialized skills that you can just talk to simply in one of your messaging apps. Again, setting

it up and configuring it will be technical. There are definitely security

technical. There are definitely security risks. And you can do basically

risks. And you can do basically everything I've seen anyone do with OpenClaw using all the other tools I mentioned. But it's worth knowing it

mentioned. But it's worth knowing it exists. And once it is all set up, it

exists. And once it is all set up, it feels more intuitive because you're just chatting with it. And this may be where everything is headed because OpenClaw is at the bleeding edge right now, but this stuff will get easier and safer pretty

quickly. Now, to get to level six, the

quickly. Now, to get to level six, the starting point is claude code. Just pick

one real problem, something you'd normally spend hours on, and just try to build a solution. Don't worry about it being perfect. The goal at first is just

being perfect. The goal at first is just to feel what's possible when you stop being limited by what you already know how to make. From there, NAD is where you go when you're ready to connect everything together into longer, more

complex agent workflows. And if you're technical enough to be curious about OpenClaw, the community around it is one of the best resources. Just know what you're getting into before you give anything that kind of access to your

systems. This final level is aspirational.

Nobody's here yet. But level seven is where everything comes together. An

actual workforce of AI agents building and running your business around the clock, growing it while you focus on the things only you can do. I think it is probably possible now to start a company

that is a oneperson company that will go on to be worth like more than a billion dollars and more importantly than that deliver an amazing product and service to the world. The first true oneperson unicorn like a billion dollar company

built by a single founder has been predicted for a while. It hasn't

happened yet but hopefully someone watching this video is going to be the one who does it. If you want to go deeper into agents again there's that resource I partnered with HubSpot on in the description. We spent a lot of time

the description. We spent a lot of time on that. And if you want just the best

on that. And if you want just the best way of helping to learn with all of this stuff along the way, we have a full course platform on Futuredia with over a thousand lessons across over 30 AI courses. You'll find full learning paths

courses. You'll find full learning paths on everything from chat GPT to video generation to coding with AI and everything in between. You can get a 7-day free trial using the link in the description. or I left this on the best

description. or I left this on the best for viewer setting so YouTube really thinks you'll like watching this video