How I replaced my $36,000 assistant with ChatGPT for $20/Month (Full 7-day playbook

If you want to be an entrepreneur — not a freelancer doing everything alone — you need managerial leverage.
From 2007 through the end of 2022, that leverage for me was a flesh-and-blood personal manager: priorities, execution, follow-up, the discipline of saying “no.”

Then January 2023 arrived, and with it, AI — specifically, ChatGPT.I replaced my human manager and got more results, more speed, more quality, while saving $360,000 over the next ten years (gross: $36k/year × 10; net, after the subscription, we’re still talking about ~$357,600).

With that delta, I’m buying another real estate property and, more importantly, I’m buying mental bandwidth: the entrepreneur’s true capital.
But here’s what nobody tells you: the change wasn’t just about cost savings.
It was about radically rethinking how value is created.

The 9 AM Email That Changed Everything

October 2022.
I wake up, like every morning, at 5:00 AM.
At 9:00 AM, the first email check.
My personal manager — let’s call him Mark — had just sent his email asking for a raise.
Many years together with excellent work.
He wasn’t asking for a better opportunity.
He simply wanted to increase his compensation.

“I can’t keep up with the pace anymore,” he wrote. “Ten parallel projects, twenty critical emails a day, every decision urgent. I’m sorry.”

I had already mentioned to him that AI was imminent and he would need to evolve and become an AI manager.
But he decided differently.
He thought he could still keep up with the growing pace without AI.

And there I was, thinking: “We’ve been together for many years, but I don’t want to stop because someone decides not to face a new challenge.”
I let 48 hours pass.
Many years together isn’t nothing.

So I decided to wait until the beginning of the following year.
ChatGPT was released at the end of 2022, and I was one of the first in the world to use it.
I laughed. A lot. I realized it was a quantum leap.
Then I got emotional thinking that Mark, by refusing to learn, had sealed his fate.

So I hit the gas pedal.
Three weeks later, I had redesigned the entire operational flow.

And Mark?
He had become completely unnecessary.
Zoom meeting, the end of a years-long relationship.
No raise — it was time for termination.
I still tried to help him, explaining that this wasn’t the end of working with us, it was the end of his job everywhere he might go.

Entrepreneur ≠ Freelancer: The difference nobody really explains

The freelancer is the lone hero in the cave.
They do everything themselves: invoices, quotes, execution, customer care, marketing, sales.
Their limit is biological: the number of hours they can stay awake and lucid.

The entrepreneur, on the other hand, builds systems that produce while they think about strategy.

Translated into brutal language: the entrepreneur buys leverage.

Financial leverage (intelligent debt).
Human capital leverage (team).
Technology leverage (software, automation).

And then there’s managerial leverage — what allows you to transform chaos into output, intentions into deliverables with deadlines and quality standards.

For years, my managerial leverage was a $36,000/year only for personal manager.

Reliable, precise, tireless.

They managed priorities, translated my ideas into executable plans, did follow-ups, said “no” when I was too optimistic in communications.

Then the future knocked on the door with an indecent proposal: “What if leverage became… another form of intelligence?”

Why this AI — This specific generation of AI — Is pure managerial leverage

Forget science fiction.
I’m not touching AGI or sentient robots.

We’ll talk in 2026 when they forcefully enter our lives.

I’m talking about real, measurable, repeatable professional tasks: research, reports, memos, emails, project outlines, editorial calendars, legal drafts, preliminary data analysis.

In the 2023–2025 period, various experimental studies (not motivational blog posts, but peer-reviewed research) have shown two robust and replicable effects:

Professional writing: With ChatGPT, time to complete complex tasks drops by ~40% and quality increases by ~18%. These results were replicated on educated professionals (lawyers, consultants, managers) in incentivized tests where people had every reason to give their best.

Fortune 500 customer support: Introduction of a generative assistant increases productivity by 14% on average, with a +34% peak for the least experienced. Now published in peer-reviewed journals — not hype, but established fact.

Software development: With an AI copilot, developers complete standard tasks 55.8% faster in controlled experiments. Not “subjective perceptions” — measured times, executable code, tracked bugs.

There’s nothing magical here. It’s cognitive offloading — you shift parts of the cognitive load outside your brain, freeing high-quality attention for decisions that matter.

Scientific literature has documented this phenomenon for decades: our working memory isn’t infinite.

There’s a famous paper about the “magical number four” — we can keep about 4±1 units of complex information active simultaneously. Everything else is strain, friction, error.
AI doesn’t just make you smarter — it makes you more efficient at using the intelligence you already have.

My experiment: chronicle of an operational transformation (January 2023 — today)

Historical setup (2007–2022)

In 2007, I hire my first personal manager.
I’m 38, scaling the entrepreneurial ladder but drowning in emails, meetings, tactical decisions.
Need someone to hold the helm.

We find the right fit after three attempts.
Average cost: $36,000/year (benefits included).
For fifteen years, the model holds.
I think, the personal manager executes.
I create, they organize.
I dream, lay down the bridge plan, and they find me the engineers and send emails to my team.

Turning Point (December 2022)

“Mark” decides not to evolve.

After many years, I’m disappointed but don’t want to stop because someone decides not to learn.
I face a choice: hire someone else (with 3–6 months of onboarding) or rethink everything.
AI finally arrives in our lives.
Goodbye hiring.

Experimental Phase (January 2023)

One Saturday morning, I open ChatGPT. I start with a simple test: “Write a negotiation email to renegotiate a supplier contract.”
The result is… decent.
Nothing extraordinary but decent.
So I double down: “Now critique yourself. What are the weak points of this email?”

ChatGPT (actually Albert, the name I assign it) gives me five sensible critiques.
I adjust.
Send the email.
Two days later, the supplier accepts the new conditions.

I think: “OK. Maybe there’s something here.”
I start programming with ChatGPT helping me with code.
I manage to connect it to the internet.
I call Joyke, my wife, and tell her, “We’ve entered science fiction.”

Aggressive adoption phase (February-March 2023)

I start using ChatGPT for everything:

Prioritization: “Here are ten tasks. Order them by impact/urgency and tell me what to discard.”
Meeting prep: “Here’s the agenda. Give me three critical questions and two risks nobody’s considering.”
Follow-up: “Generate three follow-up emails for these three stakeholders, each with a different tone.”
Research: “Summarize these three white papers and give me the implications for our business model.”

Within eight weeks, I’ve built what I call the Digital Manager — a set of prompts, procedures, checklists that function like a real personal manager.

Measured results (April 2023 — Today)

Speed: “Impossible” execution times for a human when the task is repeatable and well-defined.
Example: drafting, meeting summaries, automatic follow-up with checklists.
Tasks that took 2–3 hours now take 20–30 minutes.

Quality: Not because “AI does everything better,” but because I can force multiple revisions in minutes.
Draft → critique → refine → format → fact-check.
It’s like having a team of reviewers working instantly and for free.

Breadth: I can carry forward more parallel projects without saturating mental bandwidth.
And here’s where the real competitive advantage is born: it’s not that I work “more” — it’s that my output scales without my stress scaling proportionally.

The brutal arithmetic

$36,000 × 10 years = $360,000

ChatGPT Plus: ~$20/month (standard US plan) = $240/year = $2,400 over 10 years
Net savings: ~$357,600
With that difference, over the next ten years, I can allocate capital to:

• Another real estate property (down payment, renovation, transaction costs)
• Hires in core roles (not tactical, but strategic)
• R&D on new products
• High-conversion marketing

And most importantly: operational cash flow benefits from superior productive (and qualitative) capacity, with almost zero marginal cost.

“OK, but a human manager isn’t just about productivity. There’s empathy, judgment, political tact.”

Right.
Absolutely.
And I’m not saying replace EVERYONE.

The most solid literature doesn’t say “replace everyone,” it says “complement.”
AI especially increases productivity of the least experienced and creates widespread quality standards.

It’s what scholars call the jagged frontier: some activities are already “AI-native” (writing, synthesis, preliminary analysis), others remain exquisitely human (leadership, complex negotiation, joint creativity, P&L responsibility).

The mistake is thinking in terms of 1-to-1 substitution.
You’re not replacing humans.
You’re augmenting them.

What I Replaced (Tactical Layer)

Daily Operations

• Meeting prep: agendas, key talking points, decision frameworks
• Post-meeting synthesis: action items, deadlines, owners
• Email management: drafts, responses, follow-ups
• Document creation: proposals, reports, presentations
• Calendar management: priorities, time-boxing, conflict resolution

Research & Analysis

• Market research: competitors, trends, opportunities
• Financial analysis: basic models, scenario planning, sensitivity
• Legal review: contracts, terms, red flags (still need a lawyer for final)
• Technical documentation: specs, requirements, user stories

Communication

• Internal memos: clear, structured, action-oriented
• External comms: clients, partners, investors
• Content creation: blog posts, social media, newsletters
• Translation & localization: multi-language support

Total replaced: ~80% of what a junior/mid personal manager does.

What I kept (strategic layer)

High-touch relationships

• Key client relationships: dinner, golf, “reading the room”
• Team leadership: 1-on-1s, career development, difficult conversations
• Investor relations: trust building, narrative crafting, managing expectations

Complex negotiation

• M&A discussions: multi-party, high-stakes, lots of unspoken
• Partnership structuring: strategic alliances, JVs, complex contracts
• Crisis management: when shit hits the fan, you need human judgment

Creative strategy

• Vision setting: where are we going and why?
• Culture building: what do we stand for?
• Innovation: the truly new, not iteration

Total kept: ~20% but the 20% that matters most.

The “Jagged Frontier” in practice

Picture a jagged line dividing two territories.
Above the line: tasks where AI already beats humans (or at least equals them at 1/100th the cost).
Below the line: tasks where humans still dominate.
The line isn’t straight.
It’s jagged, unpredictable, constantly shifting.

Examples above the line (AI territory)

• First draft of anything written
• Data synthesis and pattern recognition
• Process documentation
• Basic financial modeling
• Code generation (with human review)
• Translation and localization
• Meeting transcription and summary

Examples below the line (human territory)

• Reading emotional subtext in a tense negotiation
• Making judgment calls with incomplete information
• Building trust over drinks
• Recognizing when someone’s about to quit before they know it
• Deciding which rules to break
• Taking responsibility when things go wrong

The hybrid zone (where magic happens)

The real power isn’t choosing one or the other.

It’s operating in the hybrid zone — using AI for rapid iteration and scale, then applying human judgment for refinement and decision.

Example: I use ChatGPT to generate 10 strategic options for entering a new market. Then I use my 20+ years of experience to pick the two that could actually work and refine them with nuance ChatGPT would miss.

Result: 10x faster strategy development with better outcomes than either alone.

Cognitive offloading: the science nobody talks about

Our brains have roughly the same cognitive capacity as our ancestors 50,000 years ago.

Same working memory limits. Same attention constraints. Same decision fatigue.
What’s changed?
Our tools for cognitive offloading.

Writing was the first great offload — suddenly we could store information outside our heads.
Computers were the second — we could process information outside our heads.
AI is the third — we can now think outside our heads.

The working memory bottleneck

George Miller’s famous “7±2” has been revised down to “4±1” in recent research.
We can hold about 4 chunks of complex information in working memory simultaneously.

Try remembering this while reading: 7–4–9–2–8–1–5.
Now try doing strategic planning while also remembering three client deadlines, two team conflicts, and five pending decisions.
Your brain literally can’t.
That’s why we feel overwhelmed.
Not because we’re weak, but because we’re trying to use stone-age hardware for space-age demands.

Enter the external cognitive system

ChatGPT becomes your external working memory.
“Remember these seven strategic priorities while I think about something else.”
“Hold this financial model while I consider market dynamics.”
“Keep track of these fifteen action items while I focus on the big picture.”

It’s not making you lazy.
It’s making you capable of cognitive feats impossible for any individual human.

The setup: Your 7-day sprint from zero to AI-powered

Want to replicate what I did?
Here’s the exact playbook.
No fluff, no “mindset work,” just tactical execution.

DAY 1 — baseline & mindset calibration

Objective: Understand where you’re bleeding time.

Tasks:

1. Track every task for one full day (use Toggle, Clockify, or paper)
2. Categorize: Creative vs. Administrative vs. Communication
3. Identify top 3 time vampires

Time: Throughout the day + 30 min analysis

Output: Clear map of where your time goes.

DAY 2 — Communication templates

Objective: Never write the same type of email twice.

Tasks:

1. Identify your 3 most common communication types (e.g., negotiation emails, decision memos, post-meeting notes)
2. For each, create a 4-step pipeline (draft → critique → refine → QA)
3. Test with a real case

Time: 60 minutes

Output: 3 SOPs saved and tested.

DAY 3 — Calendar & priorities

Objective: Time-boxing — recurring thematic blocks in your week.

Tasks:

1. Analyze your typical week: where does time go?
2. Define 3 recurring thematic blocks (e.g., “Sales,” “Product,” “Operations”)
3. For each block, ChatGPT prepares agenda and pre-read brief

Time: 30 minutes

Output: Optimized calendar with fixed thematic blocks.

DAY 4 — Guided research with source verification

Objective: Learn to do complex research with AI without falling for hallucinations.

Tasks:

1. Choose a topic relevant to your business (e.g., competitor analysis, market trends)
2. Ask ChatGPT for a report with strict rules: cite sources, indicate confidence, flag when uncertain
3. Manually verify sources. How accurate is it?

Time: 90 minutes

Output: First verified AI-generated report. Lessons learned on where AI is reliable vs. where you need to verify.

DAY 5 — Sales enablement (outreach & follow-up)

Objective: Automate sales sequences and follow-up.

Tasks:

1. Create templates for cold outreach emails
2. Create templates for follow-up at +48h, +7 days, +30 days
3. Generate 3 versions with A/B subject lines
4. Test on 5 real prospects

Time: 60 minutes

Output: Ready-to-use sales sequences.

DAY 6 — Context library (your AI manager’s “brain”)

Objective: Give ChatGPT access to your “business context” — documents, policies, style, examples.

Tasks:

1. Collect: brand book, product offerings, case studies, pitch deck, internal policies
2. Create a shared folder (Google Drive, Notion, Dropbox)
3. Attach key documents in relevant prompts

Time: 45 minutes

Output: Organized library that ChatGPT can “read” when needed.

Pro tip: Don’t upload everything at once. Start with the 3–5 most important documents.

DAY 7 — Retro & optimization (close the loop)

Objective: Verify KPIs, remove friction, decide whether to “promote” your AI manager to full-time.

Tasks:

1. Review the week: how many deliverables did you produce? How much time did you save?
2. What worked? What didn’t?
3. Adjust prompts, templates, workflows
4. Decide: continue or return to the old way?

Time: 30 minutes

Output: Optimization plan for the next 4 weeks.

Total time invested in the first week: ~6 hours.

Expected ROI: 10–20 hours saved every subsequent week.

Do the math: 6 hours initial investment → 40–80 hours/month saved.

Payback period: One week.

So… Is the human personal manager dead?

No and yes at the same time.
Absolutely not if you also need live actions.
Example?
Going to an office, talking on the phone.
Yes, as soon as androids enter our lives.
I can already buy the first android from Walmart.
Price?
About $22,000, like a car.
I’m just waiting for the Chinese-made robots and then I’ll proceed with the purchase. (Elon, we’re waiting.)

The balance of power has changed.
Today, a single entrepreneur can get managerial leverage that previously required multiple people, and can do it for a fraction of the cost.
This means that when you hire people in the future, you’ll hire them for what only humans (for now) can do:

• Leadership: Inspire, motivate, create culture
• Relationships: Empathy, emotional intelligence, political tact
• Complex negotiation: Reading the unspoken, adapting in real-time
• Joint creativity: Brainstorming, co-creation, serendipity
• P&L responsibility: Strategic decisions with serious consequences

AI isn’t the end of management.
It’s the beginning of meta-management — managing systems that manage work.
And that, my friends, is the future.

Closing with a touch of irony for those who made it this far

Herbert Simon, Nobel Prize in Economics, reminded us back in 1971:

“In a world rich in information, the wealth of information creates a poverty of attention.”

Today that phrase is true squared.
We live in the attention economy.
Attention — not time, not money — is the scarce resource.
Every minute you waste on tactical tasks is a minute stolen from strategic thinking.
Every email you write manually is one less decision you make on something that matters.
Working “against” AI today is like fighting with the elevator by using the stairs out of moral principle.

Yes, you’re exercising.
Yes, you’re burning calories.
But you arrive late.
And in business, arriving late is losing.

When your business is running well, you can afford to climb the stairs just for exercise.

The real point (what nobody says out loud)

It’s not about “AI vs. humans.”
It’s about augmented humans vs. unaugmented humans.
And in five or at most ten years, the difference between these two groups will be abysmal.

Those who embraced technological leverage — AI, automation, systems — will have built empires with reduced teams and enormous margins.
Those who resisted on “principle” will still be there, heroically, doing everything themselves.
Tired, burned out, and obsolete.
And want to know the hard truth?
90% of them won’t exist anymore.

Think about how many professions and businesses you’ve seen disappear since 1995.
Now the same thing is about to happen multiplied by a million.
It’s not a moral question.
It’s a question of strategic choice.
If you want to be an entrepreneur, buy leverage.

And then use that leverage to buy the most undervalued thing of all:
Peace of mind.
The peace of mind knowing your business runs even when you’re not working.

The peace of mind to say “no” without guilt.
The peace of mind to sleep eight hours without anxiety.
That, my friends, is priceless.

But it only costs $20 a month.

PS — I don’t take commissions from Sam Altman