- A consultant trained on slides, not real deployments.
- A pure technologist coding without watching ROI.
- An AI commentator without a system in production at a company to back it up.
Hi, I'm Alex.
Entrepreneur since 2017. AI-First architect since 2025. I help companies of 20 to 500 employees rebuild their organization around AI. Rare triple profile: entrepreneur, tech, and business.
AI isn't a layer you add — it's an organization you rebuild. I build AI-First organizations: a competitive edge that compounds every month and very quickly becomes uncatchable.
- 2017First bet
- 20M EURBootstrapped revenue
- 4Companies founded
- 95Agents in production
Positioning
There are two camps in AI in France.
I'm in neither.
On one side, consultants who theorize without ever shipping anything. On the other, pure technologists who code without understanding why. The profile that does both — entrepreneur and AI agent architect in production — is extremely rare.
- An entrepreneur with a business view who lived the transformation across 4 companies.
- An architect who personally designs the company's knowledge architecture and agents.
- An entrepreneur with 9 years of P&L responsibility and 20M EUR of bootstrapped revenue behind him.
14 years, 4 bets, one common thread.
Never wait for the perfect moment. Test. Pivot. Measure. Repeat.
- 012012ENGINE
Rio, age 22
The non-negotiable reason
One night in Rio, I run for my life, chased by two armed men. I escape by jumping into a taxi. Since then, I know everything can stop in a second. That's what pushes me to act without ever waiting for the perfect moment.
- 022017CHAPTER 1
Live Escape Box
1,000 EUR and 36 calls
I launch my first company with 1,000 EUR. To get into FNAC, we call Katia Azoulay 36 times. She had blocked our number. Today the boxes are in every FNAC store. Smartbox (400M EUR revenue) copies the concept. We fight. We win.
- 032019-2023CHAPTERS 2-3-4
Boomrang + Albus Factory
Going to scale
30,000 participants/year. Covid pivot decided in 15 minutes, 2,800 digital events delivered. Then Albus Factory: 60 custom immersive experiences for LVMH, Apple, Amazon, BCG, Capgemini, Naval Group, TotalEnergies. 20M EUR cumulative revenue. 100% bootstrapped.
- 042025THE PAUSE
The year out of time
Understanding from the inside
I step out of operations for a year. Two GMs take over, zero contact. 6 months in Asia — Thailand then Bali. I build an AI model from scratch and a custom RAG memory, autonomous agents, to understand how it works from the inside. As a bonus, I become a semi-pro Padel player.
- 052026THE PROOF
The system
Proof in production
Back in Paris in February. After 8 weeks of sprint, I've shipped 95 agents across my companies, 11 plugins, 150+ memory files. 9 people now do the work of 50. Albus Factory becomes the living proof of the thesis.
Since 2017
Boomrang Events and Albus Factory have worked with these organizations.
These references come from my events and serious-game work, not yet from my AI practice. The AI proof is Albus Factory itself.
The stories I tell in meetings.
N° 012017
36 calls to get into FNAC.
Katia Azoulay, head of FNAC gift boxes, had blocked our number. We kept calling. Thirty-six times. She finally picked up. Today, Live Escape Box gift boxes are in every FNAC store in France. Lesson: when you have no network, persistence is a competitive advantage.
N° 022020
Covid pivot in 15 minutes.
70% of revenue gone overnight. I gathered the team on a video call. In 15 minutes, we decided to pivot everything to digital events. Two weeks later: site live, 30 vendors lined up, first events with BNP, Amazon, Capgemini. 2,800 digital events delivered during the crisis.
N° 032025
6 months in Asia. The system ran without me.
I spent 6 months in Asia. Thailand, then Bali. Two GMs running the companies. Zero contact with the teams. I wasn't on vacation, I was building. An AI model from scratch. Then a RAG memory system: an LLM that remembers everything, session after session. The problem everyone hit and nobody truly solved. Back in February 2026: 95 agents shipped in 6 weeks. The 'year out of time' was the foundation.
N° 042026
The 2 SEO agents that changed everything.
I built my first team on Claude Code — two agents, two weeks. One analyzes keywords, the other writes. The moment I got it: this isn't a tool, it's a team that keeps context and improves with every run. A month later, 100+ agents in production.
N° 05Imperfect
The agent that almost sent client pricing.
One of my early prospecting agents sent a reference email that included exact sale prices: 'SNCF ordered a Serious Game for 23,200 EUR for…'. Fixed in 30 minutes. An agent does exactly what you tell it. Nothing more, nothing less. Supervision stays non-negotiable. That's the difference between a toy and a system.
N° 062026
The agent that killed the 5,000 EUR/month SEO agency.
Albus Factory was paying an SEO agency 5,000 EUR per month. Decent results, high price, total dependency. I plugged Claude Code into our tools and set up cron tasks. In two weeks, the agent replaced the agency. It works 10× better. 60,000 EUR saved per year. Zero marginal cost after that. That's when I understood I had to rebuild everything around agents, not just bolt an AI layer on top.
What I believe.
What I prove.
90% of companies use AI as a gadget. A chatbot here, a summary there. What changes everything is to think the company around AI. Not adding: rebuilding.
- 01
AI absorbs the repetitive. Humans handle the strategic.
Agents take over the mechanical work. Humans focus on strategy, relationships, complex decisions.
- 02
Triple profile or nothing.
Three skills that rarely coexist: business judgment, technical depth, operational experience. Most players master only one. Those who master all three are rare. It's the non-negotiable condition for a real transformation.
- 03
ROI comes from restructuring.
Adding ChatGPT to a bad organization = a slightly better bad organization. The gain isn't in the tool. It's in redesigning workflows around agents. Measurable in months, not years.
- 04
Agents that learn change everything.
A static agent is a tool. An agent with persistent memory that learns from its mistakes and from new data is a skill that compounds. It gets better every week without you touching it. That's the real revolution — not automation.
- 05
The gap widens fast. And doesn't close.
An AI-First organization improves every week with no marginal cost. A traditional one doesn't. In 2-3 years, the gap becomes exponential. Today's pioneers have 18 months of lead. That kind of lead doesn't get recovered.
Self-taught.
I break the status quo.
Zero technical training. From Bubble in 2017 to autonomous agents in 2026, I learned everything in the heat of the action. I don't fit any box of standard AI consulting — and that's exactly what makes the difference.
- The profile
- Builder, not manager
A good day is when I've built something. An agent that runs, a system that works, a tool that solves a real problem. Not a well-run meeting.
- The drive
- Fireworks of ideas
A 'Yes' profile in the Richard Branson sense. I generate more ideas than I can execute. I surround myself with 'But' profiles who challenge them.
- The format
- Sprints, not steady flow
When I'm in a topic, I'm fully in it. 95 agents in 6 weeks is a total sprint. Then I decompress. Fixed schedules kill creativity.
- The filter
- Indifferent to optics
I don't seek external validation. My freedom and enjoyment matter more than image. That indifference isn't a pose — it's what lets me hold sharp opinions.
If you've read this far, you're already in the right quartile of leaders.
The rest will keep reading LinkedIn threads. You took the time to understand who's behind the proposal. That's the same instinct that separates SMEs that transform now from those who'll watch others do it eighteen months from now.
- Alex