Founder of Meetme and Baboom

Romain de Wolff

Feeling incredibly grateful today 🙏

I had the privilege of publishing my very first opinion piece in a Swiss magazine, Bilan.

Seeing my words shared in a national newspaper, online, feels both surreal and deeply humbling.

This first publication is a small step on paper, but a huge one for me personally. It represents years of learning, experimenting, questioning, building, and above all, being surrounded by amazing people who supported, challenged, and trusted me along the way.

The article explores how AI assisted dictation is bringing lawyers back to the essence of their craft, voice, reasoning, intention, and freeing time for what truly matters. Writing about this shift, and seeing it resonate in such a respected Swiss publication, is a real honor.

A big thank you to the Bilan Magazine editorial team for the trust and the opportunity.

And thank you to everyone who believed in me, directly or indirectly, this would not have been possible without you.

Looking forward to writing and sharing much more. This is just the beginning ✨

Article: https://lnkd.in/dqQg2ffs - in french 🇫🇷🇨🇭

Open article: Avocats: la dictée assistée par IA révolutionne le métier | Bilan by bilan.ch

Avocats: la dictée assistée par IA révolutionne le métier | Bilan

Over the last 18 months, I have spent a lot of time onboarding lawyers to AI.

Hundreds of sessions, across firms, teams, cultures, sometimes very skeptical rooms.

What always strikes me is that the real challenge is not adoption.
It is understanding what AI actually is, and what it is not.

AI is not reasoning in the human sense.
At its core, it is statistics applied to language, predicting the next word based on probability.

Once lawyers get this, many things suddenly make sense.

Why early models were terrible at simple math problems.
Why accuracy degrades when you feed too much information at once.
Why context size matters far more than people think.

Give AI too much, entire case files, thousands of pages, and quality drops.
Not because it is lazy, but because of how it computes.

The best results come from restraint.
Carefully selected documents.
Clear questions.
Intentional scope.

The same logic applies to hallucinations, especially in legal matters.

When you ask an AI to recall a specific case or ruling from its training data, it may sound convincing and still be wrong.
Not maliciously, just statistically.

This is why legal research tools that first retrieve, then reason, change everything.
They ground the model in real sources before it speaks.

The lawyers who get the most value from AI are not the ones who trust it blindly.
They are the ones who constantly challenge it.

AI does not replace legal judgment.
It sharpens it, if used with discipline.

I keep seeing the same pattern.
Understanding the limitations is what unlocks real efficiency.

Curious to see how this collective understanding will evolve as models improve, and as our habits adapt with them.

Picture: in a law firm office in Geneva, during a training

No alternative text description for this image

Lean AI transformation, without heavy IT changes for SMEs and Investment firms.

We help mid-market companies deploy custom AI agents and train teams for adoption, delivering real results in weeks. See detailed case studies at voilalabs.ai.

How we help
- Audit & Discovery: we map your data sources & processes, uncover high-impact AI use cases through focused interviews with your teams and prioritize them in terms of ROI
- Build & Deployment: we deploy custom AI agents seamlessly integrated with your systems and data sources; system is secure by design, with no heavy IT changes required
- Training: we upskill your teams for lasting adoption & impact: covering AI fundamentals to hands-on use of your custom AI agents; we also train power users to maintain the system and build new agents as your needs evolve

Why clients work with us
We keep it practical and lean, focused on measurable outcomes and quick wins that bring ROI.
From a customer: “Great technical knowledge, pragmatic business acumen, very good collaboration.”

Next step
- Book a 30-min discovery call (Calendly link in profile).
- Prefer email or DM? max@voilalabs.ai or DM me here.

Maxence Jouve

Leaving a top-tier consulting firm to start my own business was one of my best decisions last year.
Here are my top 5 decisions, business / career-wise of 2025:

1) I took the leap

I had dreamed about entrepreneurship since I was a teenager.
This was the first time I actually did it.

What I love: ownership, controlling my time, and working only on what matters.
No "bullshit" tasks to justify downtime. No politics either.

2) I found the right co-founder

Stan and I met in 2016 (first year of uni). Back then it was ML and deep learning, but we were already obsessed with where AI could go.

Being aligned on mission and working style makes everything easier.
It also makes hard weeks more bearable.

3) I committed to a space moving fast

When I was younger, I dreamed about being at the start of the internet era (The Social Network was one my favorite movie). Everything was still to be built.

AI feels like that now.
Exciting. 
Sometimes terrifying, with releases that make you rethink everything overnight.

But that volatility is exactly why the window is now open.

4) I am focusing my work in the real economy

AI consulting keeps me close to the real economy: operations, teams, and messy IT systems that actually run companies.

We are not at the edge of AI development, but our work is grounding and make us realize companies have their own pace going through changes (and it slower than what you see online).

5) I am learning constantly, very broad skills

Going independent forces you to do everything yourself. And you realize you have blind spots.

Turns out being extroverted ≠ being good at sales.

Since starting, I upskilled myself in building offers, running discovery calls, prospecting, selling, coding production systems again… Skills I never touched at BCG.

AI Engineering @ Voila AI | SMEs & Investment firms | ex-BCG | EPFL

One of my 2026 resolutions: learning German (from A1 to B1 😅).
This morning, AI saved me 3 hours of prep work. Here's how:

I am taking online classes for grammar and conjugation, but learning a language also means memorizing lots of vocabulary.

For that I use Anki flashcards on my phone (it shows a word in German or English, and you guess the translation).
It is the app used by med students to optimize knowledge retention.

I can then learn vocab while moving around Zurich, in the tram.
However, the bottleneck is getting high-quality cards into the app.

Anki uses .apkg files. Not exactly a common format.
Building custom decks manually is hard. I don’t even know how to do it.

This morning, after my lesson on irregular verbs, I gave the class pdf to Claude Cowork.
I also included in one sample .apkg file (to show format).

Five minutes later: I had a perfectly formatted deck, ready for my phone.

Cowork doesn't just chat. It executes tasks by clarifying objectives with you, reading your local files, and creating final outputs.

A year ago, I wouldn't have even tried it.

Now I can create a custom deck whenever I want to learn something specific.

Takeaway: AI is evolving from a conversational tool into a true executing assistant.

taplio

The Battle

Romain de Wolff

Maxence Jouve

Followers

Posts

Engagements

Likes

Performance ScoreRomain de Wolff's LinkedIn Performance

Performance ScoreMaxence Jouve's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Romain de Wolff

Maxence Jouve

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Romain de Wolff's LinkedIn Performance

Performance ScoreMaxence Jouve's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreRomain de Wolff's LinkedIn Performance