Madeleine Labs · Est. 2010
Madeleine Labs. Go Experience-Powered
Tribal experience is your most precious asset. We extract your company's best judgment and make sure it runs everywhere.
My name is Nicolas Doumenc and since 2010, I've worked with 300+ experts and I've put their experience to work on dozens of problems (see a sample below).
I named my “lab” Madeleine (after the famous “madeleine de Proust”) and I take on projects that push the limits of what's possible with experience extraction and injection, often using AI as a catalyst.
“Nicolas's passion is studying concepts that will catch people off guard. And it works! When I invited him to a CAC 40 executive committee meeting to play the wild card, the CEO picked up one of his concepts to conclude on the group's 4-year strategy.
“I worked with Nicolas on preparing a transformation project using the military tactical reasoning method. I loved the format: very fast work phases where you get the concept and move forward immediately. Within the first half hour, we completely reframed our objective and I had real insights that will help us succeed.
“When Nicolas proposed extracting our experience, we were skeptical: how do you capture more than 20 years of extreme expeditions and research in just a few sessions? Three weeks later, our team uses a tool that retrieves the right field anecdotes for each client, helps us create commercial proposals aligned with our DNA, and frees up time for us to focus on our scientific expeditions and our client expeditions.
Veolia - Antifragile Workshop
I was part of the strategy offsite for Veolia's executive committee: my role was to roam the room to inject concepts and challenge ideas. I wrote a brief about Antifragility for the CEO between day 1 and day 2 of the workshop. The next morning, Antoine Frérot and I discussed it. He then grabbed a whiteboard, wrote "Veolia Antifragile," and built his closing address around it.
VEOLIA 5 Year Plan
Worked directly with Veolia's executive committee on their 5-year strategy plan. Challenged the plan using survivorship bias and risk management frameworks.
T'choupi for the C-Suite
A parody of France's most beloved children's activity book, reworked as a strategic planning workbook for Veolia's executive committee. I designed it as a pre-read before the strategy offsite: wild cards, lean startup, superstar effects... Some execs scoffed; the Secretary General, CEO and COO, walked into the room with their copies, ready to work.
Martin Serralta pushed the AI clone concept to its limit: 30 questions to help you find your life's momentum, used by over 1,000 people, generating more than 100,000 messages. This isn't an experiment anymore, it's knowledge transmission at scale. The cherry on top: despite the volume, qualitative feedback was outstanding, with many users writing to the real Martin to say his digital twin made them cry.
Madeleine on WhatsApp
Created the Madeleine AI Agent: every morning she pings you on WhatsApp to extract a piece of your experience from the day before, 10 years ago... Then you can ask her questions and see how different people on your team would solve a problem.
“I've worked with tons of consultants in my career and I thought I knew strategy pretty well. Yet I spent 2 hours with Nicolas and the team of military officers he'd invited, and it increased my project's ambition tenfold.
“The first meeting with Nicolas hit me like a ton of bricks. He's the only guy I've met who pushes back on me like that. Too often I throw out an idea to the group, and since nobody challenges me, we end up with a shaky system. Not so with Nicolas. He has a level of standards and truly radical ideas that force me to work and rework topics I thought I had mastered. As a result, I'm making progress like I rarely have before.
AI Agent for 30 sales
Extracted a CEO's gut feeling through military debriefs and simulations, turned it into an AI agent for 30 salespeople. They actually use it: 90% adoption, faster deal velocity.
Bid intelligence, not bid automation
A global consulting firm contacted me because they failed at building custom AI tools for RFPs. I showed their team that the real leverage isn't automating the process, it's extracting expert insights to find the winning angle. Simple tools loaded with the right experience beat any custom platform, because you can't automate a process no one can explain clearly to a human.