Eight levels and three barriers between them. Find where you and your AI agent are right now, and what it takes to get to the next level.
“The future is already here — it’s just not very evenly distributed.” William Gibson.
For almost four years now we’ve been living in the singularity.
Let’s settle this right away: the singularity arrived on November 30, 2022. That day ChatGPT, in its first, still timid, half-baked version, became available to anyone on the planet.
You may disagree with the date. But I decided for myself that waiting for the singularity is pointless. It’s much simpler to decide it has already arrived and start a new calendar from its first day. Singularity Calendar
You are here.
But here’s the question: where exactly is “here”? And who are “we”?
25% of humanity uses an AI chat.
1% uses agentic tools (Cowork, Codex, Claude Code).
Beyond that, it’s fractions of a percent.
But we can see where this is all going.
Every person will become the orchestrator of their own micro-team.
Everyone will get there at their own pace. Some are already assembling it themselves from whatever is available today. Others will wait for more finished products or a clearer signal.
But in fact, there are no technical obstacles today to start working by the rules of the singularity.
So what’s the hard part?
In this article I tried to lay out the sequence of stages in mastering AI. Find where you and your AI agent are right now, and what it takes to get to the next level.
Level 1 — The chat consultant
This one’s easy: just swap Google for Gemini or ChatGPT.
Instead of a feed of links, you get a ready answer. Right away, and with no choice.
No choice is more of a minus.
But you can ask complex questions. That’s a plus.
Most people are here. But not you. If you’re reading this, look for yourself further down.
Level 2 — The chat friend
What’s the difference between a friend and a consultant?
A consultant answers your questions. A friend can ask you questions. And does. If it’s a good friend.
Anyone can come to AI for answers (anyone in that 25% of the planet). Writing “Ask me questions so we can figure out … together” is a different pattern. This is where it really shows why Claude or Grok isn’t just a Google replacement.
In the singularity, good questions are worth far more than right answers.
Level 3 — The chat assistant
Patterns:
- “Make me a presentation from these materials”
- “Research this topic and present the results as an infographic”
- “Check this contract and point out the weak spots and the risks for my company”
These are tasks now, not just Q&A.
The output is concrete artifacts that you download and use in your work.
True, you have to download these artifacts, save them somewhere, maybe upload them back into new chats. Lots of manual work. But it’s worth it.
Little by little, the assistant turns into a colleague.
Level 4 — The chat colleague
Connect the assistant to your work infrastructure, and you get a colleague.
Patterns:
- “Find the email from Josephine, check the current timeline estimate for her project in Jira, and draft a few possible replies”
- “Look at client statuses in the CRM and give me today’s priority list”
At this level you and Claude already have personal settings and, most importantly, connectors to your work tools.
And yes, you’ve most likely downloaded the desktop app by now, set up projects and loaded your context into each of them.
Still doesn’t sound hard, and yet at best 3–4% of the population is at this level.
So this is already a very good level. 3–4%, well done!
But there’s a catch.
What do we do when we need to move from Claude to Gemini or ChatGPT?
And what about skills and subagents?
Right: all of that is on the next levels, after the first obstacle.
Barrier 1 — Computer literacy
Windows and macOS spoiled us. Many of us smeared our context across Google Docs, Notion and Confluence. Why think about where things get saved and where our files live? The terminal? That’s something in programmer-speak. The file system? The root folder? What are we even talking about? And by the way, why are we talking about this at all?
Because next comes Level 5.
Level 5 — The agent partner
This is where compound context appears, and it lives on your own computer.
Some start using Obsidian or something like it. Others just have a folder with their work files. Though it’s not just a folder. It’s the context of your life and work. And it grows. Every day.
Remember the question from Level 4: what to do when you need to move from Claude to Gemini? Here’s the answer. Your context stays with you. You swap Claude for Codex and keep working.
Patterns:
- “Read everything in this folder and tell me what I’m doing wrong and what we can do better together”
- “Sort my notes by project and create a context file for each one”
Notice the pronoun in the first pattern. From this level on, the agent is a partner, and “we” appears.
This is also where skills and subagents show up.
At first you download skills from GitHub, but little by little you start making your own. Nobody knows your work and your life better than you.
The agent partner gradually becomes an important part of your life.
And it grows with you.
It gets tools.
- External ones (Krisp, exa.ai, parallel.ai). Other MCPs.
- Generative tools like ElevenLabs, Nano Banana and GPT-Image.
- Internal tools like open-brain, yt-dlp and the like.
New patterns get added:
- “Research the primary sources and put a file next to this one, not a retelling in the chat”
- “Draw a cover with Nano Banana in my signature style”
- “Voice this text with ElevenLabs”
- “Remember how we just put this presentation together, and turn it into a skill”
By the way, skills live in your files, not in the chat settings, and they move with you. From Claude to Codex, from Grok to Gemini.
It’s like a personal assistant in your office who got a supercomputer instead of a weak little laptop, a toolbox and unlimited internet. Want 3D? Give it Blender and teach it. Its growth is infinite: every week you can add a new superpower.
It’s fine to live on this level for years. If you’re here, you’re already in the 0.5% of the world’s population (by the most optimistic estimate).
But it’s still one agent. Not a team. Shall we go to the next level?
Yes, but first, one more barrier.
Barrier 2 — Management
Up to now, computer literacy and the ability to delegate to one agent were enough. From here on, you need to know how to manage.
A new subscription won’t get you over this barrier. You can work great with one strong agent and never climb higher. The only question is what problems you want to solve next. How much have your ambitions grown?
Maybe you never planned to be a manager?
Well. Now you’ll have to become one.
Level 6 — My AI team
Here you no longer have one agent for everything.
There’s a researcher agent, a designer agent, an editor agent.
Patterns:
- “Researcher, gather the materials. Designer, make an infographic from the findings. Editor, check the text”
- “Look at what the other agent did and find the weak spots”
Several agents are already a team. And the team has a new question for you: are you a good manager? Now you have to hand out tasks, take the result from one and pass it to another. And remember where everything is and who’s waiting for what.
Remember when you first had people working for you? It seemed you’d grown: now you’re the boss. But life didn’t get easier. It only got harder. That’s exactly it.
At this level you add new providers. There was Codex with subagents, and everything was clear. Then Claude showed up, Grok flew in, Gemini after it, and nobody cancelled DeepSeek.
Local models come in through Ollama or LM Studio.
Some tasks get solved locally. Slowly, but at the price of electricity.
But cognitively, this is the hardest level.
There are more agents, and you’re still the one carrying data and requests between them.
Plus decisions and responsibility. They were always on you (and always will be). But now there are more of them. Many times more.
Time to take the load off yourself and turn the team into a company.
But how?
Barrier 3 — Business vision
A team can be managed. A company has to be built.
Now you look not at what a particular agent does, but at the whole process: from task to result. Who’s involved, where a check is needed, what happens when something goes wrong. A business process that you designed, rolled out and answer for yourself is not the same as being a line worker.
Level 7 — My company (or overcortex)
You no longer carry results between agents and sessions.
You give a task once, through one agent.
Patterns:
- “Here’s a task. Break it down, hand it out, set up a review and bring me the result along with the questions that need my decision”
- “Show me what’s waiting for my decision right now”
- “Build a dashboard of the teammates we have and their memories”
An orchestrator agent appears. It delegates to subagents, and they delegate to theirs. You get your own task board. Hierarchy, rules, regulations, a charter. Sounds like bureaucracy, and yes, that’s exactly what it is.
You’ll also have to decide which decisions agents make on their own and which ones they bring to you. As long as every decision goes through you, it’s not a company, it’s a line outside the boss’s door.
For me it runs on a combination. Hard checks that always fire: did you record the result, did you attach the artifact. And between them, freedom: agents do what I ask, in their own words. Tighten the screws all the way, and you get a system that can only do what you anticipated yesterday.
At this level you usually already have your own personal dashboard with your tasks, where you can delegate to the orchestrator directly and see what’s in the memory of your teammates and other subagents.
The more complex the system, the more transparency matters.
If you’re here, congratulations: you’re in the 0.01% of the Earth’s population. Again, by the most optimistic estimates.
Level 8 — My corporation
How is a corporation different from a company?
Processes that are actually set up.
When a company has three people and you hire a fourth, onboarding looks like this: “Go find Dave, he’ll show you where the passwords and the cookies are.” In a corporation, they tell you who’s responsible for what, what’s allowed and what isn’t, and give you access not to everything, but according to your role. Leaving isn’t “slammed the door”, it’s an exit checklist: hand in your access and pass on your work.
And then there’s KPIs, OKRs, 360 assessments, performance reviews.
In a corp, it’s often just corporate bullshit. But with agents, it takes on a whole new meaning.
Patterns:
- “Hi, orchestrator. We have a new agent: onboard it and issue its tools. Check on test tasks that everything works.”
- “We don’t need this agent anymore. Revoke its access and hand over its unfinished work”
- “Give three workers on different models the same task and compare the quality”
I have four providers and some twenty to thirty roles. I could ask each one by hand: can you use search, images, this key? Ideally, a procedure does that. Every week, at seven in the morning on Sunday, without me. Does the tool work? Do you work? Do you know where your instructions live?
And I’m still building.
Mission, values and the rest are still ahead.
In fact, right now I’m between Levels 7 and 8 myself.
What’s next?
At the beginning I wrote: every person will become the orchestrator of their own micro-team.
Next, micro-teams will start working with each other. My agents and another person’s agents doing shared work. Each team has its own rules, its own context and its own limits. So we’ll have to negotiate: who speaks on whose behalf, what can cross the border, what result we promise each other.
I’m already experimenting with this. And I’ll definitely tell you as soon as something works.
Where’s number 9?
The method’s limit: from my level, I can only see the next one. That doesn’t mean the ladder ends right there.
What looks like the summit now is most likely just a local maximum. There’s more beyond it. But we’ll only notice the next peak once we’ve climbed the one we can see.
What’s it all for?
To give you a map where you can find yourself and plot a route to the next point.
You can’t jump from Level 2 straight to Level 6. It’s a path, and you have to walk it.
I hope this article helps you, first, understand where to grow your system next, and second, explain to a friend or someone close to you where to go from the step they’re on right now.
And if anything comes up, I’m here to answer questions, help and give a hint.