This post was originally made in French, which is one of my native languages.


Take Cuban Salsa dance. The music sets the tempo, one, two, three... five, six, seven.
Keep your basic step and you stay in rhythm. You can add figures, improvise, play with your partner. But the moment you get excited and throw too many moves without feeling the music, you think you’re doing “more” when really you’re losing the beat. You pull your partner off-rhythm and the whole dance becomes uncomfortable.

I learned this from a dear friend who teaches dance. She showed me that rhythm isn’t just technique, it’s about being “human”. In dance, like music, you have to feel the other person, listen, breathe, stay present by connecting with your partner and environment, like tuning into a radio frequency. You can be technically brilliant in your steps, but if you lose that presence, you lose the dance.

That lesson stuck with me, especially now when technology moves faster than we do. We quickly forget that without human sensitivity, every system eventually falls out of sync, even with the best tech tools. That’s the part I keep in mind when working with innovation, especially AI.

When I started working with AI agents, which are becoming mainstream in 2025, automating almost everything we see on screens, I realized most problems come from exactly the same place. People want to “do more” for various reasons, but no longer feel the rhythm of their own system. They forget certain human principles.

With AI today, most teams or individuals live exactly like this. On paper, AI saves you 60-70% time on certain tasks. But in real teams, part of that gain gets eaten up by reviewing, checking, confusing instructions, judgment errors, and unlearning.

We talk a lot about quick wins with over-automation of repetitive tasks, but we often forget the losses, like the 10% of time spent on control, the errors forcing you to redo work, the 56% of users admitting they’ve made mistakes because of AI, and the 66% who don’t even always verify the responses they receive.* (sources in footnotes)

Result? Just like in Cuban Salsa: 3 steps forward, 2 steps backwards.
They’re moving forward, but desynchronized from their own system.

The problem isn’t the dance, the music, or even the advanced moves.
The problem is adding complexity without a stable base, without clear internal rhythm. Without mastering fundamentals, it’s like putting the cart before the horse, or a baby learning to run before knowing how to walk.

What Musashi Theorized 400 Years Ago, a Bit of Japanese Philosophy.

In a Substack post in French that inspired me to write what you’re reading, Oussama Ammar mentions Age of Empires, a game from my youth that I used to enjoy playing. His analogy immediately reminded me of a book: The Book of Five Rings, written by Miyamoto Musashi, the best swordsman of his era in Japan about 4 centuries ago, reputed to be unbeatable. Back then, losing a duel meant losing your life on the spot. No second chances. It helps to picture the comfort we have today to test ideas without paying the ultimate price.

In this book, Musashi explains a simple concept that transcends all eras:
“Rhythm”. And his rule is clear:

“He who imposes rhythm commands, he who suffers rhythm loses.”

In the video game Age of Empires, that’s exactly what happens. You don’t win because you have more units, you’re constantly creating. You win because you control the invisible tempo of the game. The civilization that progresses isn’t the one clicking the most, but the one transforming each click into a system. You structure your economy, you define your rhythm, and everything follows. The army and wealth are just side effects of the rhythm you’ve built.

For samurai in Japan, the katana wasn’t “just another tool.” It was an extension of the arm and intention. They said the sword had a soul because it entered symbiosis with whoever wielded it.

Without a body, it served no purpose.

Without spirit, it didn’t cut at the right moment.

AI is the same logic. It’s not an entity replacing humans, it’s an extension of their field of action. It doesn’t think for us, it amplifies what our internal rhythm gives it to process.

The real leverage today isn’t “how many tasks are automated?” but

“at what rhythm does your system breathe?”

A human capable of creating their AI agents, documenting their workflow, and stabilizing their tempo is worth far more than a manager spending their week in useless meetings, convinced they’re “busy,” perhaps trying through that show, like so many others with their phones, to maintain the illusion of their utility to society or their organization. A very ephemeral status, with what’s coming in the future.

The 3 Strategic Rhythms (Or How Not to Get Eaten by Your Own System)

To scale your projects, whether solo or leading 50 people, you must coordinate 3 rhythms simultaneously:

1. Internal rhythm → How you structure your execution. Your personal work cadence or your team’s. It’s your “basic step” in salsa or any martial art or expertise. If you don’t master it, everything else is fragile. So you can already forget the shortcuts you glance at on social media with the hype around certain influencers, with their 5 bullshit tips (copy-pasted from ChatGPT) to succeed at... (AI Slop!)

2. External rhythm → How you read market cycles, customers, constraints. You can’t dance alone (a partner or partners could be nice), and you need to FEEL (I said FEEL with human feeling) the music around you. A 2-week internal sprint, ignoring a 3-month client cycle, is guaranteed to cause desynchronization.

3. Adaptive rhythm → Your ability to change tempo without breaking your system. Musashi insists on this:

“Find your own rhythm, impose it, then change it before the other understands.”

In business, that means knowing how to go from “exploration” mode to “execution” mode without losing your teams along the way.

Questions to Diagnose Your Own Rhythm

Here are the questions I systematically ask (no universal right answers, just tensions to identify):

On your internal rhythm:

  • How many times per week do you feel “overwhelmed” by your own tools or AI agents?
  • Which loop (extraction, analysis, decision, execution) takes 80% of your time currently?
  • If you had to “slow down” one of the 3 loops to regain control, which would it be?

On your external rhythm:

  • Does your internal sprint (2 weeks, 1 month?) match the real cycle of your clients or market?
  • How many times do you adjust your roadmap or calendar because of external signals you hadn’t anticipated?

On your adaptive rhythm:

  • When you change priorities or strategy, how long does it take your team (or agents) to adapt?
  • Do you have an “emergency mode” that breaks your system, or does your system absorb emergencies?

On your polyrhythm (if team):

  • Do your different teams run at different speeds? If yes, where do they intersect?
  • When a project involves 3 teams, who decides the final tempo?

These questions don’t give you answers directly if you’re in a hurry.
They reveal where your system dances off-beat.

And remember this well: there are only good questions, never ready-made good answers. There’s no fire. Unless the equation is truly impossible, the right answers, if they don’t come now, sometimes end up appearing with time. Because the first step of progress is asking them.

These 3 strategic rhythms aren’t theoretical. When you implement AI in a team or project, you’ll immediately see where they’re desynchronized.

The 4 Tactical Loops (The Engine Running Under the Hood)

You don’t have “one single work rhythm.” You’re running at least 4 loops, each with its own speed:

1. Extraction loop → Everything you capture as input: notes, screenshots, transcripts, audio, video, newsletters, journals, raw signals.

It’s fast, continuous, almost automatic.

2. Analysis loop → What your brain or AI agents transform into ideas, structures, decisions, or possible options. This requires distance, detachment, stepping out of the action to catch your breath and gain perspective.
Musashi recommended training in nature to see your blind spots.

3. Decision loop → The moment you cut like a katana. You say yes, no, or later (Reflect). It’s slow, human, irreplaceable. (Human-in-the-loop)

4. Execution loop → What actually comes out into the world as output: emails, offers, content, posts, products. This can be fast (AI agents) or slow (humans), depending on what you’ve built and how you document your processes before automating them.

These loops don’t all land at the same moment. Like in music, they overlap, offset, then fall together on key points. If you try to align everything to the same tempo, you end up burned out. If you let each loop go off in its own direction without a framework, you end up in chaos.

The real game is to keep these 4 loops alive inside your 3 rhythms.

How It Scales from Solo Founder to 50+ Person Company

Here’s where it gets interesting and where most frameworks fall apart.

Solo founder: You’re the only conductor of your own rhythms.
You extract, analyze, decide, and execute. Your AI agents helping in the mission mainly amplify your execution loop (and sometimes analysis). It’s manageable as long as you document your “basic step” well (Notion.com for example).

Team of 5-10: You start multiplying rhythms. Each person has their own internal tempo. Your job becomes synchronizing decision points without killing autonomy. AI agents help everyone manage their own loops, but you must define where loops intersect (stand-ups, reviews, deadlines).

Team of 20-50: Here, you can no longer impose ONE single rhythm. You build an intentional multi-rhythm system. Different teams run at different speeds (product ≠ sales ≠ support ≠ HR), but they must fall together on key points every month or quarter (launches, board meetings, KPIs, OKRs).

The trap at this stage: taking shortcuts and wanting to “unify” all rhythms!
It’s impossible and counterproductive. The product team needs 2-week sprints, sales needs monthly cycles, support runs continuously.

The real skill of scaling a business isn’t imposing a single tempo, it’s coordinating different rhythms without friction.

AI agents, in this context, become “translators” between rhythms.

An agent that extracts customer insights for sales, transforms them into structured feedback for product, and keeps the support team in the loop... that’s orchestration. It’s teamwork! And if you don’t know how to work as a team with humans, even if you claim otherwise, with AI it won’t work either.

Musashi would say: You don’t change each musician’s rhythm, you change when they play together.

Concrete Example: The Classic Desynchronization

Their problem wasn’t AI. Their problem was rhythm.

I saw this 2 months ago with a friend’s SME, 35 people in the creative industry.

In 4 weeks, their visible output dropped 60%, but their impact tripled. Less noise, more relevant signal. Their posts finally generated conversations... and leads with sales conversions. The goal wasn’t just generating dopamine with useless views (for ego) but actually invoicing with money hitting the bank account. On social media, views aren’t synonymous with sales. The game has changed.

The diagnosis was simple: broken multi-rhythm.

Too much information extraction, too much execution, zero time to analyze and decide properly. The agents were running faster than humans could think.

What I Actually Do (No BS)

We just celebrated ChatGPT’s 3rd birthday (November 30, 2022).

Since its release, I update myself daily while considering myself a lifelong student, and I implement AI in varied contexts, from Fortune 500 corporate enterprises like HP Inc., where I led multilingual AI projects, to small startups. What I’ve learned:

Tools change, but the rules of the game stay the same.
Rhythm has always been the foundation.

My approach isn’t to “deploy AI agents” because the latest AI tools are cool or just “train people on how to use these AI tools.” It’s to:

  1. Map the current 4 loops (where it’s blocked, where it’s overflowing).
  2. Identify rhythm desynchronizations (teams stepping on each other, lack of leadership, human inefficiencies, decisions taking weeks because too many intermediaries or nobody knows who decides).
  3. Build a system that breathes, not one that sprints permanently, but stays in constant motion, even slowly like a tortoise. As long as it keeps functioning. And if we can’t, we do it in our heads planning the next decision. It’s a process to be appreciated at each step rather than lived at full speed.

I don’t believe in speed without knowing where to focus. I apply mastered rhythm.
Not too slow, not too fast! Just the right rhythm to savor the moment and instant. As Musashi wrote 400 years ago:

“If you don’t understand your rhythm and the adversary’s, you cannot win.”

And today, with AI as a natural human extension, our cognitive extensions force us to understand this seriously.

If we want to adapt, there are no escape routes or shortcuts with magical AI solutions, as those Hype influencers sell just to get views and grab your attention. Artificial intelligence is anything but magical.

AI doesn’t replace your rhythm. It exposes every desynchronization.
And for better or worse, it amplifies it.

Alors on danse.

Notes (sources)

(1) MIT/Wellesley, Your Brain on ChatGPT (2025) and Stanford, World Bank study show 60-70% gains on certain tasks.

(2) Multiple 2024, 2025 analyses estimate knowledge workers spend an average of 4.3 hours per week verifying AI outputs, just over 10% of their time.

(3) KPMG and University of Melbourne, Trust, attitudes and use of AI, Global Study 2025: 56% of employees report having made a mistake due to AI, 66% don’t verify response accuracy.