Written by a manager who couldn't code. Still can't.
I'm not a programmer.
I'm not an AI expert.
I'm not the guy building the models everyone talks about.
I'm a manager.
Just like you.
And last October I got a phone call that should have been exciting.
I was getting promoted.
Instead of celebrating, I opened my calendar.
Because the promotion didn't come with more time.
Or more people.
The work was about to grow.
My team wasn't.
I had six months before the new role started, and one question that wouldn't leave me alone:
How am I supposed to do all of this?
At first, I wasn't looking for AI.
I was looking for time.
I needed to get dramatically faster at everything I was already doing.
And if I'm honest…
I wanted something more than that.
There were projects I'd wanted to tackle for years but never had the bandwidth.
For me, that was building marketing support for our dealer network.
So I started learning.
Every morning.
Three o'clock.
Coffee.
Laptop.
A few quiet hours before breakfast with my daughters, the one appointment in my calendar that never moves.
Then straight into a workday that usually ends closer to eight than five.
I did that for months.
This page is everything that came out of those mornings.
So you don't have to spend six months figuring it out yourself.
Before we go any further...
Let's make sure this is actually for you.
This is for you if...
Your company introduced AI, but nobody really showed you how to use it in your actual job.
You've opened ChatGPT, asked a few questions, and thought, "I could've written this myself."
Everyone seems to be "using AI," and you're quietly wondering whether you're already behind.
Your week is full of work that feels repetitive, but there's nobody to delegate it to.
You'd genuinely like to learn AI... but evenings belong to your family, not another online course.
This probably isn't for you if...
You're a developer.
You're looking for a collection of clever prompts.
You enjoy learning about AI more than actually using it.
If you're still reading...
You're probably exactly where I was.
Here's what nobody tells managers.
Maybe your company never trained you.
Maybe they did.
But it was built for everyone.
Which usually means...
...it wasn't built for anyone.
So you go looking on YouTube.
LinkedIn.
Courses.
Newsletters.
Reddit.
Everyone has a different method.
Everyone promises shortcuts.
Everything feels technical.
And underneath all of it sits one question:
Even if I learn all of this... will it actually work for MY job?
So you close another tab.
Monday arrives.
The same report takes the same three hours it took last month.
I know.
Because that's exactly where I lived.
The idea that changed everything
For almost a year I consumed AI content.
Videos.
Podcasts.
Courses.
Notes.
Always thinking I was one tutorial away from finally starting.
Then that promotion happened.
And suddenly I couldn't afford to stay in "learning mode."
That's when something obvious hit me.
Managers spend their entire careers managing people.
We brief people.
Train people.
Correct people.
Delegate to people.
We already know how to build high-performing colleagues.
So why was I treating AI like software...
...instead of treating it like a new employee?
That single question changed everything.
Instead of learning prompts...
I started managing AI.
I gave it responsibilities.
Processes.
Context.
Feedback.
Standards.
Exactly the way I'd onboard someone joining my team.
That became Bruno.
My AI colleague.
(Yes, he's named after the character from the movie my daughters insisted on watching over and over.)
Then another realization made everything simpler.
Everyone says AI needs complicated systems.
Maybe.
But not to get started.
Bruno lives inside something every manager already understands.
Folders.
That's it.
The same folders you've been organizing work in for years.
Instead of learning a completely new way of working...
...you build your AI inside the system you already know.
Which makes it surprisingly easy to understand.
What happened next surprised me.
Reports that used to take hours now take minutes.
Alongside my full-time job, I now run a marketing operation that produces the work of a three-person team.
Eventually the system got enough attention that I found myself presenting it to senior global leadership.
Not because I'm an AI expert.
Because I'm a manager who built something practical.
And it worked.
Here's what we're going to build together.
Over the next few days...
You'll build your own Bruno.
Not another chatbot.
Not another demo.
A real AI colleague.
One that understands your work.
Knows your files.
Follows your processes.
And takes over one real part of your job.
Exactly the way you want it.
Here's the roadmap.
Here's my offer.
Fifteen minutes a day.
That's it.
No coding.
You can even start from your phone.
And one promise.
By the end, you'll have an AI colleague that understands your work, follows your process, and handles one real task exactly the way you want it done.
Systems build results.
My guarantee
It's the only guarantee I can honestly make.
If you spend fifteen minutes a day following along...
...and by the end Bruno isn't handling one real part of your work...
Send me a message on LinkedIn.
I'll personally help you get there.
Ready?
Filippo
Day 1: AI doesn't know you. Today we fix that.
Fifteen minutes, any chat window, even from your phone.
Welcome to Day 1.
Look at the map.
Problem number one:
AI doesn't know YOU or YOUR job.
That's why every answer feels generic.
Not because you're bad at "prompting."
Because the model has never met you.
Think about the smartest consultant in the world, walking into your company this morning.
Brilliant. Expensive.
Useless.
Not because he lacks intelligence.
Because nobody has briefed him yet.
That's been your AI, all along.
So today we flip the whole thing around.
Today, AI interviews YOU.
Your fifteen minutes
🟨 TOOL TODAYAny AI chat you already have. ChatGPT, Claude, Gemini, it doesn't matter. Even on your phone, even on the couch.
Here's how it works.
You paste one text.
The AI starts asking you questions.
One at a time.
About your role, your work, and the task that eats your week.
You just answer.
You're describing your own job.
And nobody on earth knows it better than you.
At the end, it hands you two things.
One: your briefing. A short document about you, your work, and your rules. Save it anywhere you'll find it tomorrow: your notes app, a Word file, an email to yourself. It's about to become the first thing your Bruno ever reads.
Two: your task. The ONE piece of your job Bruno takes over this week.
Here's the text. Copy it, paste it, answer honestly:
COPY & PASTE
"I'm a manager building an AI colleague for my work. Context for you, so you never need to ask me about it: this interview will produce a briefing. Tomorrow that briefing becomes the permanent memory file (CLAUDE.md) of an AI assistant living in a folder on my computer, and later this week it becomes the base for teaching that assistant to run one of my recurring tasks (a skill). You don't need to explain any of this to me.
Your job today: interview me, one question at a time, in plain language. No jargon, no technical questions, and never ask me why I'm doing this or what tool I'll use: you already know. Find out: my role, what my work looks like, how I like things done, and the ONE task I should delegate first. The task must pass five checks: I repeat it every week or month · it's built from files I already have on my computer · I can show a finished example of a good version · the result is a file of any kind (a report, a spreadsheet, a presentation, a written document) · it needs no live data from the internet. If my first idea fails a check, tell me why in one plain sentence and help me pick a better one.
When you have everything, write "BRUNO'S BRIEFING v1" as plain text I can copy and save anywhere: who I am, how I work, my rules, and the chosen task. No file formats, no technical vocabulary inside it. Then stop."
When the briefing is done, add one line to it yourself.
By hand.
The most valuable line I own:
"If you don't know something, write UNKNOWN. Never guess."
Here's why.
AI wants to please you.
And a colleague who wants to please you will invent an answer before admitting he doesn't know.
You've met humans like this.
Now yours has the rule in writing.
Forever.
⚠ Honest warning: the interview will feel almost too simple. That's normal, answering is the whole job today. Keep your answers short and real. And if the AI starts lecturing instead of asking, just tell it: "one question at a time."
✓ You did it if: you have a briefing document with your rules inside, and ONE chosen task that passed the five checks.
Congratulations.
Most people never get past asking AI random questions.
You just gave it a proper briefing, like a manager onboarding a new hire.
Problem 1 is already cracking.
HOW DID IT GO?
Everything smooth, briefing saved.
Got there, but something felt off.
Stuck.
Whatever your color, send me a DM on LinkedIn: DAY 1 + your color.
Green, I celebrate with you. Yellow or red, I get you unstuck.
I read everything.
Tomorrow, Bruno gets a body.
You'll build his home in about ten minutes, put your briefing inside...
...and watch him read YOUR files back to you, by name.
See you tomorrow.
Filippo
Day 2: AI can't see your files. Until today.
Ten minutes of setup, then he reads them back to you. By name.
Yesterday, an AI interviewed you.
You walked away with a briefing and one chosen task.
Today, that briefing gets a reader.
Look at the map.
Problem number two:
AI can't see your files.
You know this pain already.
Uploading attachments into a chat, one by one.
Re-explaining the same context, every single time.
And some files you'd never upload anywhere anyway.
Here's the fix, and it's almost embarrassing how simple it is.
You don't bring your files to the AI.
You bring the AI to your files.
Bruno is going to live inside one folder on your computer.
Not your email.
Not your desktop.
Not your company systems.
One folder.
You decide what goes in, so you decide what he sees.
That folder is his body, and you're about to build it.
Your fifteen minutes
🟨 TOOL TODAYVS Code, a free program by Microsoft, plus the Claude Code extension inside it. This is Bruno's home. After today you'll open it like you open Outlook.
Honest heads-up before we start.
Today is the heaviest fifteen minutes of the week.
Installing, signing in, one password.
Boring, I know.
But you only do it once, and everything after today is the fun part.
One practical note first.
Some company laptops are locked down: nothing installs without IT's blessing.
If that's you, don't fight it.
Build Bruno on your personal computer.
He's YOUR colleague, not your company's.
(Just pick task files you're comfortable keeping there. If your chosen task can't leave the office, pick your second-best task for this week and keep the first for later.)
Here we go.
Build the body. On your desktop, create a folder with your name on it. Mine is filippo.ai. Inside it, create one folder for the task you chose yesterday, and inside that: a folder called input (the files you use to build it), a folder called example (one good finished version), and a folder called output (empty, for now). Drag the real files in.
Install VS Code. Download it, install it like any program.
Open your folder. In VS Code: File, then Open Folder, then pick your folder. Now look left. Your work, sitting in a sidebar, exactly like folders in Outlook. Nothing on that screen is new. It's your email client wearing different clothes.
Install the brain. Click the blocks icon on the left. That's the extensions store. Search for Claude Code, install it, click the Claude icon that appears, and sign in.
⚠ Honest warning #1: this is the one thing in the whole system that costs money. About 20 dollars a month for the starter plan. My math: one lunch, for a colleague who works nights. It's the only cost in everything I'll ever show you.
5. Wake him up.
Open the Claude panel, and paste this, together with your briefing from yesterday:
COPY & PASTE
"Below is my briefing, produced yesterday by an interview with an AI assistant. I'm a manager with no technical background, so work in plain language and never ask me technical questions. Do this, in order.
One: read the briefing, then look at every folder and file in this directory. If anything important is unclear, ask me, one question at a time.
Two: create a file called CLAUDE.md, exactly that name, in the main folder. This file is your permanent memory: you read it automatically at the start of every session. Write into it my briefing, my rules, my chosen task, and what you learned from my folders, so that you always know who I am and how I want things done. Create no other file, and don't change any of my existing files.
Three: when CLAUDE.md exists, tell me what you see in this folder, in your own words: my files, by name, and what you understand my task to be.
[PASTE YOUR BRIEFING BELOW]"
One thing before he starts.
Bruno asks permission before reading or changing anything.
A small box will pop up: Yes, or Allow.
That's a safety feature, not a problem.
He's a polite colleague. Click yes.
Now watch the screen.
He reads.
He asks a question or two.
Then he names YOUR report.
YOUR files.
YOUR task.
Back to you, in plain sentences.
The first time it happens, it's strange.
In the best possible way.
⚠ Honest warning #2: VS Code looks like a developer tool, because it is one. You'll see buttons you don't understand. Ignore 90% of them. The intimidation wears off in about ten minutes.
✓ You did it if: there's a new file in your sidebar called CLAUDE.md that Bruno wrote himself, and he described your files back to you, by name.
(Curious about the .md ending? It's just the text format all AI tools read best. Click it: VS Code shows it to you like a normal document. Word, for colleagues with silicon brains.)
And about his memory.
Every time Bruno wakes up, the first thing he reads is that file.
Your world. Your rules. Your task.
Every session, automatically, forever.
(Memory that carries whole conversations across days? That comes later in this journey. For now, the file IS the memory, and it's all we need.)
One more thing, because you should know what you just bought into.
Nothing.
The brain inside Bruno is swappable.
Today it's Claude, because I use it, and because right now it's the best.
Tomorrow it could be Codex, from OpenAI.
Or Gemini, from Google.
Or one day a local brain that runs entirely on your own computer, so nothing private ever leaves it.
Here's the point.
You built the body. The folders. The briefing. The rules.
That part is yours, forever.
The brain, you simply plug in: always the best one available, swapped the day something better ships.
The smartest companies on earth are spending billions competing to improve your colleague's brain.
We don't compete with them.
We build on top of them.
Congratulations.
Your AI has a body, a brain, and a memory file he wrote himself.
Problem 2 is gone: he reads your world, and only the part of it you gave him.
Bruno is alive.
HOW DID IT GO?
Bruno read my files back to me.
Installed, but something felt off.
Stuck somewhere.
DM me on LinkedIn: DAY 2 + your color.
Green, I celebrate with you. Yellow or red, I get you unstuck. Setup problems are the most common ones, and the fastest to fix.
I read everything.
Tomorrow, the hours start coming back.
You'll teach Bruno your task the way you'd teach a new hire...
...and by the end, it becomes a button.
See you tomorrow.
Filippo
Day 3: The handover you wish you'd gotten.
Five steps, and your task becomes a button.
Look at the map.
The last problem standing:
AI won't do YOUR work consistently.
Brilliant on Tuesday.
Different on Thursday.
And a colleague you can't rely on isn't a colleague.
Before we fix it, a memory.
At some point in your career, you got the bad handover.
A laptop.
A phone.
And a "good luck, kid."
Nobody explained the reports. Nobody showed you what good looked like. You figured it out.
But it took months longer than it should have.
Today we do the opposite.
Today you give Bruno the handover you wish you'd gotten.
The way you'd want to be managed.
And here's the good news: this is a five-step framework, and you already know how to run every step.
The Handover: five steps
🟨 TOOL TODAYNone. Today's tool is a manager skill you've had for years: a proper handover. Bruno just receives it faster than any new hire you've ever had.
Step 1. Context first.
Open Bruno and paste:
COPY & PASTE
"We're starting the handover of my chosen task, the one in your memory file. Before anything else, get your context: re-read your memory file, then study the finished version in `example` and every file in `input`. If any file won't open for you, tell me which one. Then summarize, in plain language, what you understand so far and what you still don't know."
Notice something while he reads.
Yesterday's conversation is gone.
Bruno doesn't remember chats. He remembers his memory file and your folders.
That's exactly why we built them.
And if you ever want his eyes on one specific file, don't describe it.
Drag it straight into the chat.
No uploads. No attachments. He's already home.
Remember this rule, it's worth more than every prompt trick ever sold:
The right context beats the perfect prompt. Every time.
Our folders make context a drag and a drop. (One detail for VS Code: hold Shift while you drag a file into the chat.)
Step 2. Show the dish.
Think of a chef.
Before anyone cooks, the chef shows a photo of the finished dish.
Your example is that photo.
Drag it into the chat (remember: hold Shift).
Then tell Bruno, in your own words, what makes it good: the structure, the tone, what the reader needs to see first, what must never be missing.
Two or three sentences. Your words.
Step 3. Give him the recipe.
The dish is the output. The ingredients are in input.
Drag those in too.
Now the recipe: every action, in order, exactly as you'd teach it.
Where each number comes from. What gets checked against what. What you round, what you never round. What you do when something's missing.
The more detail, the better.
You're not writing code. You're explaining your Tuesday to a colleague.
Tell him: "Ask me questions until the recipe is complete."
Answer until his questions run out.
Step 4. First run together, correct the output.
Tell him:
COPY & PASTE
"Now produce the task, following the recipe exactly. Write the finished work only into `output`, nowhere else. Then tell me you're done."
Open the result. Compare it to your example.
Something will be off. Good.
Correct it, and always say WHY.
Not "fix the intro."
"Fix the intro: my director reads only the first three lines, the conclusion goes there."
The why is what turns a fix into a rule.
Run again. Correct again.
Until the output is genuinely yours.
Step 5. Bruno runs the task alone.
One last test, the one every real handover ends with.
COPY & PASTE
"Run the task one more time, start to finish, completely on your own. No questions this time. Then tell me what you did and where the result is."
Open it.
If it's right, and it will be, the handover is complete.
He learned your task. Not a task. Yours.
The handover is done.
All five steps.
Now put the framework down for a second.
Because what comes next is where I had my first wow moment with AI.
The moment I understood what this actually was.
And the moment I deliver on the promise I made you at the start.
Let's create a skill.
Now make this output a permanent skill.
Paste this, exactly:
COPY & PASTE
"The handover is complete and this output is the standard. Now create a skill from it, in the official Claude Code skill format, so you load it automatically in every future session. Do exactly this. One: pick a short, simple name for the skill based on the task, lowercase, words separated by hyphens. Two: create the file .claude/skills/<that-name>/SKILL.md inside this main folder, creating any missing folders. Three: start the file with the required frontmatter, name: (the skill's name) and description: (one sentence: what this skill produces and when to use it). Four: below that, write the complete procedure in plain language: my recipe, every correction I made and every "why" I gave you, where input and example live, and the rule that finished work goes only into output. Change nothing else in my folders. Five: verify your own work: read the file back, confirm the frontmatter is valid and that this skill will load by itself in a brand-new session, and fix it if not. When everything checks out, tell me three things: the skill's name, the exact words I type to run it next time (the slash command, plus one plain sentence that also works), and the file's location."
He writes it down.
He hands you the words.
Write them somewhere you'll see them.
A sticky note on the monitor is allowed. Encouraged, actually.
Because the task that eats your week...
...is now a button.
⚠ Honest warning: the first run will be about 80% right, and the missing 20% will be things YOU forgot to put in the recipe. That's not failure, that's the most honest mirror you've ever had. And yes, today can stretch past fifteen minutes. It's the deepest day of the week, and the one that pays for all the others.
✓ You did it if: the solo run produced your task correctly in output, a skill exists with your recipe and corrections inside, Bruno confirmed it will load on its own in a new session, and the run words are on a sticky note in front of you.
Congratulations.
Today you did the thing almost nobody does.
You didn't use AI. You onboarded it, corrected its thinking, and delegated a real piece of your job.
The handover you wish you'd gotten: you just gave it.
HOW DID IT GO?
The button exists and the solo run looked like mine.
It ran, but it's not quite my standard yet.
Stuck.
DM me on LinkedIn: DAY 3 + your color.
Green, I celebrate with you. Yellow or red, I get you unstuck. Yellow today usually means one more correction round, and that's the system working.
I read everything.
Tomorrow, my favorite day.
You'll press your new button.
Then I'll hand you one question managers almost never think to ask...
...and your report will come back better than the one you've been sending for years.
See you tomorrow.
Filippo
Day 4: Press the button. Then ask the question.
The one question managers almost never ask.
Before anything else today, look at this chart.
Each dot is millions of people
2,500 dots · color = most advanced AI interaction
Never used AI · 84%Free chatbot · 16%Pays for AI · ~0.3%Runs a real system · ~0.04%
based on estimates shared by Alex Finn, Feb 2026
Every dot is millions of people.
Grey: never touched AI.
Green: free chatbot users.
Orange: the few who pay for it.
Red: the tiny group running AI on their own files, with their own system.
Four days ago, you were green.
Today you're in orange, and after yesterday, honestly?
You have one foot in red.
You can now run a real piece of your job with a single word.
(Go ahead, run it again right now. You earned it.)
So today, let's see what you can do that the others can't yet.
The Round Table
I imagine AI as a round table.
Albert Einstein is sitting there.
Elon Musk.
Miles Davis.
Mike Tyson.
And any world-class expert, in any discipline you can name.
That's Bruno.
Bruno is any of them, whenever you need one.
Now, here's the thing about that table.
It's the same table for everyone. Same experts, same seats.
What changes is YOU.
Most people ask generic questions.
Would you really ask Einstein how the weather looks?
Or they ask the wrong expert.
Would you ask Elon Musk for marriage advice?
(I love Elon... but I would not.)
Not you. Not anymore.
Your Bruno knows you and your job.
Full context, your standard, and a real task he already runs on your behalf.
Which means the experts at your table can finally advise you like the world-class professionals they are.
Let's test this.
Run your skill. Your words, press enter.
(Notice something: this is a brand-new session. Yesterday's chat is gone. Bruno read his memory file, found your skill on his own, and ran it anyway. That's the system proving itself.)
And while the fresh output sits in front of you, ask the one question managers almost never think to ask:
"What could you do to make this better?"
Step one: ask the table.
Paste this:
COPY & PASTE
"Look at the finished work in `output`. You know my task, my standard, and who receives this work. Now review it like a table of the best professionals in the field: what could make it more valuable to the person who receives it? Give me your best ideas in plain language: what you'd change, why it's better for my reader, and what you'd need from me. Ideas only. Don't change anything yet."
Now read the list.
Somewhere in it, there's an idea that makes you stop.
Something useful you never thought to ask for.
On your real work.
That's the table earning its seat.
Step two: pick like a manager.
Don't take everything.
Ideas are cheap now. Judgment is the expensive part, and the judgment is yours.
Pick one or two that genuinely serve your reader.
Ignore the rest without guilt.
Step three: demand a plan.
Here's a rule you already live by.
Nobody changes a working process without showing you a plan first.
Your report works. So Bruno doesn't touch it until you've approved his thinking.
🟨 TOOL TODAYPlan Mode. At the bottom of Bruno's chat box in VS Code there's a small mode selector: click it and choose Plan. From that moment, Bruno must lay out his plan and wait for your approval before touching anything. The video shows the exact spot. And if your screen looks different — these tools update often — just ask Bruno where Plan Mode lives. He knows his own house.
And a money reason to love it.
AI work is billed in tokens: every word read and written costs a little.
Work done wrong and redone costs three times the tokens, and three times your patience.
Correcting a plan costs almost nothing.
You already manage people this way. Now you manage Bruno this way.
Switch Plan Mode on, and paste:
COPY & PASTE
"Let's apply the improvements I chose: [WRITE THE ONE OR TWO YOU PICKED]. Propose your plan, step by step, in plain language, and wait for my approval. After I approve, apply them and save the improved version in `output`, next to the current one. Change nothing else."
Read the plan. Correct the thinking if something's off. Approve.
Step four: open both versions.
The one you've been sending for years.
And the new one.
Look at them side by side.
That difference is what the table is for.
Step five: make the upgrade permanent.
Paste this:
COPY & PASTE
"The improved version is the new standard. Update the skill file you created yesterday inside .claude/skills: add the improvements we chose and the reasons behind them, so every future run produces this new version. Keep the skill's name and its frontmatter valid. Change nothing else in my folders. Confirm when it's done, and tell me if my run words stay the same."
Done.
Your button now produces the better version.
Every time.
⚠ Honest warning: the table will happily generate twenty ideas, and taking too many is the real trap. A report that tries to say everything says nothing. One or two improvements per round. You can always hold another round table next month.
✓ You did it if: you saw v1 and v2 side by side, the skill now carries the improvements, and your run words still work.
Congratulations.
Look at the map: Problem 3 is gone.
Your task runs consistently, at a standard that just got higher than the one you've been sending for years.
And you did it the manager way: ideas from the table, judgment from you.
HOW DID IT GO?
v2 exists and it's genuinely better.
Improvements came out messy or too big.
Stuck.
DM me on LinkedIn: DAY 4 + your color.
Green, I celebrate with you. Yellow or red, I get you unstuck. Yellow today usually means the improvement was too ambitious: shrink it, rerun.
I read everything.
Tomorrow, the last day.
No new installs, no new tools.
Just three things: the full list of what you actually built this week, one last gift...
...and where this road goes next.
Bring a coffee.
Filippo
Day 5: Graduation day.
Count what you built this week.
Five days ago, you had a chat window.
Generic answers. Uploads. Starting over every morning.
Now count with me what's sitting on your computer right now.
A briefing about you and your job, written by an AI that interviewed you.
A body: your work, organized in folders he can see.
A brain: the best available today, swappable the day something better ships.
A memory file he re-reads every single morning, with your rules inside.
A skill: your task, your recipe, your corrections, running on a single word.
And a report that comes out better than the one you've been sending for years.
That's Bruno.
Body, brain, memory, one real piece of your job.
The promise from day zero. Kept.
And one more thing, the part I care about most.
You now own the method.
The interview. The handover. The round table.
They work for ANY task.
Your second skill will take half the time you spent on the first.
Your third, even less.
That's the part nobody can take from you.
Remember the chart?
Yesterday you had one foot in red.
Look again today.
Both feet.
You're in the group the chart can barely draw.
One last gift.
Managers live in three documents.
Excel. Word. PowerPoint.
Anthropic, the company behind Claude, publishes official skills for all three, free, in one pack.
(And to be clear: nothing you built this week needs them. Your skill already runs. This is an upgrade, not a repair.)
One install, and Bruno stops answering in walls of text.
He hands you a real spreadsheet. A real document. A real deck.
Type these two lines into Bruno's chat box, one at a time, enter after each:
(These two start with a slash because they're commands for Bruno's home, not sentences for Bruno. That's normal. If anything asks you to confirm or restart, say yes.)
Then paste this:
COPY & PASTE
"I just installed Anthropic's official document skills (Word, Excel, PowerPoint, PDF). Confirm they're ready to use. If anything went wrong with the installation, guide me through fixing it step by step, in plain language, one step at a time. Then give me one example of something I can now ask you to produce from my own task files. And remember from now on: when I ask for a document, hand me a real file, not a wall of text."
(Curious where the pack lives? github.com/anthropics/skills — Anthropic's public library. Yes, there's a fourth skill in the pack, for PDFs. Consider it a bonus gift.)
You will thank me later.
Where this road goes.
Now the honest part.
Bruno today is a capable junior.
Sharp, reliable, trained on your world.
And I know exactly what your next weeks with him look like, because I lived them.
Three new walls are coming.
Wall one: every output still needs your eyes.
He works, you check. Everything, every time.
Real delegation starts when he checks his own work, and you review exceptions, not everything.
Wall two: he runs one task. Your job is whole processes.
Collect, build, check, send. And he still has no arms: he can't reach the web, your email, your other tools.
That was on purpose. First week, safety first.
Wall three: he works only when you do.
Close the laptop, and the operation stops.
Your time is still the bottleneck.
I hit all three walls. I broke all three.
Here's the map for that road:
At the end of it, the math finally works:
A bigger job, in the same hours, with Bruno running half of it.
That's the next part of this training, and I'm building it as you read this.
Free, like everything so far. One email a week.
Want it? Leave your email here, and you'll get it the day it's ready: