Clef de voûte

Vous vous demandez où en est votre orga produit & Tech sur l'IA ? Passez notre AI maturity check ici (3 min)
Reçois les analyses approfondies de chaque épisode en t'inscrivant à la newsletter Clef de voûte
--
Pennylane, c'est aujourd'hui une licorne française valorisée 3,5 milliards d'euros, 115 millions d'euros d'ARR, 800 000 entreprises clientes et 1100 salariés.
Et comme toute boîte de cette taille, l'enjeu du moment est le même : comment on structure une équipe Product autour de l'IA, sans brusquer les experts et sans casser ce qui marche déjà ?

Pour en parler, j'ai reçu Tancrède Besnard, cofondateur et CPO de Pennylane.
On avait déjà enregistré un épisode ensemble il y a trois ans.
Depuis, tout a basculé.

Chez Pennylane, la transformation IA passe par des choses très concrètes.
Une équipe dédiée de 6 personnes qui construit le tooling IA en interne.
Un bot nommé Jean-Claude qui exécute du code assigné à un ticket Linear.
Un framework pour cadrer l'adoption dans les équipes.
Et une conviction forte : pas un licenciement au nom de l'IA.

Si tu es CPO, VP Product, fondateur ou VP Engineering et que tu te demandes comment déployer l'IA dans ta boîte sans brusquer ton équipe, cet épisode est pour toi.

Quelques-unes des questions abordées :
Comment structurer une équipe Product de +100 personnes autour de l'IA ?
Où se déplace le vrai goulot d'étranglement quand l'IA écrit le code ?
Comment convaincre les experts sceptiques sans les brusquer ?
Quels outils utiliser pour Product, Design et Engineering aujourd'hui ?
Le conseil de Tancrède aux CPOs qui démarrent leur transformation IA

Bonne écoute.

---
[00:00] Introduction
[02:00] Rappel de Pennylane et son ambition IA
[07:59] La chaîne de production Product en 4 étapes
[15:24] Optimiser le lien entre les étapes
[18:03] L'équipe Apex et Jean-Claude
[22:46] Pas de suppression de rôles chez Pennylane
[25:30] Où se déplace le nouveau goulot d'étranglement
[27:20] Le vrai enjeu du PM à l'ère IA
[29:44] Convaincre les experts sceptiques
[33:48] Le framework Adopters / Explorers
[35:36] Mesurer l'adoption et les Tips of the day
[38:55] La stack outillage Product & Engineering
[40:06] Le PM qui utilise Cursor sur la codebase
[44:38] Figma vs Lovable : deux usages
[50:28] Office hours et communication
[54:19] Apprentissage contre-intuitif
[55:29] L'échec : vouloir aller trop vite
[56:23] Le conseil pour un CPO qui démarre
[59:20] Comment Pennylane recrute avec l'IA
---
💥 Pour apporter ton soutien au podcast : 
1. Abonne-toi pour ne rien manquer 🔔
2. Laisse un super avis sur Apple podcast ou Spotify ❤️ 
3. Rejoins la chaîne Youtube

Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

Wondering where your Product & Engineering organization stands on AI? Take our AI Maturity Check here (3 min).
Get the in-depth analysis from every episode by subscribing to the Clef de voûte newsletter.
--
8 months ago, OpenAI invited me to San Francisco to shoot a first documentary on how Silicon Valley startups use AI to build their products.
This time, Figma invited my Co-Founder David and me.

So we spent 3 days at Config, Figma's yearly gathering in San Francisco. 10,000+ designers from the biggest tech companies came to see what's next.
Our question, the same as last time but more focused: what is AI really doing to the designer's job?

Over 1 week, we interviewed more than 30 designers from very different companies, right on the floor of Config.

We also had the chance to interview the VP Product and the VP Design at Figma.
We attended the keynote from Dylan Field, Figma's CEO and watched Figma announce Code Layers, a launch that redraws the line between design and code.
What we're bringing back: some designers are more excited than ever.
Others are clearly worried.
No one gives the same answer to the same question.

A few of the topics we dig into:
How designers are actually experiencing the AI shift
What Figma is preparing next with Code Layers
Where the real value of a designer sits now
What AI can never take from a designer (and why nobody agrees)
How to design products for AI agents

Enjoy.
---
[00:00] Introduction : Back to SF
[01:38] Arriving at Config
[02:40] The Figma keynote and Code Layers
[05:57] What AI is really changing for designers
[07:32] When the design/code wall comes down
[10:04] When the lines between roles blur
[13:26] What AI can never take from a designer
[18:36] What we're bringing back from San Francisco
---
💥 To support the podcast:
1. Subscribe so you don’t miss an episode 🔔
2. Leave a great review on Apple Podcasts or Spotify ❤️
3. Join the YouTube channel

Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

Wondering where your Product & Engineering organization stands on AI? Take our AI Maturity Check here (3 min).
Get the in-depth analysis from every episode by subscribing to the Clef de voûte newsletter.
--
In this episode, I have the pleasure of welcoming Nicolas Dessaigne, Partner at Y Combinator, one of the most influential startup accelerators in the world. Nicolas is a repeat founder who co-founded Algolia, a company that became one of Europe’s biggest B2B SaaS successes before expanding globally. Today, he supports hundreds of founders as part of YC, helping shape the next generation of iconic startups. We recorded this conversation in San Francisco, where Nicolas shared his unique perspective on how YC has evolved — from 2 batches a year to 4 — and how this new AI era is transforming the way startups are built. We talk about the new wave of AI-native founders, why product iteration speed matters more than ever, and how large companies are struggling to keep up with the pace of change. Nicolas also shares what makes YC truly unique, the role of technical founders, and why code is no longer the moat it used to be. If you’re a founder, a product builder, or just curious about how YC sees the future of startups in the AI era, this episode will give you a rare inside look.
---
[00:00] Intro
[01:36] YC in 2025
[02:37] Batch model & investment strategy
[03:50] Billion-dollar success rates
[05:14] Alumni support post-YC
[06:44] What makes YC unique
[08:25] Partners are ex-founders
[09:20] New AI product categories
[11:13] AI in healthcare, robotics
[12:27] His product preferences
[14:03] Moonshot companies & YC goals
[16:00] Why some repeat YC
[17:11] Post-YC momentum as a predictor
[18:21] Founding with a family
[19:49] AI native vs integrated products
[20:30] AI as standard infrastructure
[22:11] Founders are getting younger
[24:43] How YC startups build fast
[26:52] Balancing speed vs quality
[29:39] Pressure creates creative founders
[31:25] Code is no longer precious IP
[33:11] Spring 2025 batch growth stats
[36:08] Rapid AI tool evolution
[38:00] AI vs humans in support
[39:46] Case study: Prosper
[40:54] Evals as a key asset
[43:15] AI amplifies all founder types
[44:41] AI’s effect on larger orgs
[46:54] Risk of inertia in scale-ups
[47:20] Future of PM and eng roles
[49:04] Tools lower the bar to entry
[50:47] How non-AI YC alumni adapt
[53:14] Pre-AI companies
[55:11] Examples: Replit, Vercel
[56:55] What Nicolas would work on today
[58:56] Backing founders
[59:28] Robotics vs. Progress
---
💥 To support the podcast:
1. Subscribe so you don’t miss an episode 🔔
2. Leave a great review on Apple Podcasts or Spotify ❤️
3. Join the YouTube channel

Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

Wondering where your Product & Engineering organization stands on AI? Take our AI Maturity Check here (3 min).
Get the in-depth analysis from every episode by subscribing to the Clef de voûte newsletter.
--
In this episode, I’m really excited to welcome Spenser Skates on Clef de Voûte.
Spenser is the co-founder and CEO of Amplitude, the digital analytics platform that helps more than 4,500 companies around the world understand their users and build digital products that matter.
In our conversation, Spenser will share what it really takes to grow from the scrappy early days to meaningful enterprise scale : the trade-offs, the tough choices, and how you keep moving fast as the company gets bigger. We’ll also talk about building real product differentiation with AI : how to see through the hype, spot what’s real, and focus on long-term value rather than chasing trends. And finally, Spenser will give us a look at Amplitude’s roadmap for AI : where the company is headed next, how they’re planning to deploy AI, and how it will positively impact their customers.
---
🔨 4 pillars to structure a top product-tech team
🔨 How to launch a market-ready product in 18 months ?
🔨 The method to turn a technical tool into a self-serve SaaS
🔨 How to structure your roadmap in 2 layers (6 months / 6 weeks)
---
[00:00] Intro
[02:40] Origin of Amplitude
[06:15] First customers
[09:30] Product analytics market
[14:01] Why not compete with GA
[16:50] Building a self-serve product
[18:50] User friction and onboarding
[20:15] Feedback loop with support
[23:00] Product org
[25:30] PMs and data modeling
[28:00] Structuring tech & product teams
[31:45] Vision vs execution cycles
[34:20] Product verticals
[38:00] UI for different industries
[41:10] Amplitude’s competitors
[44:00] What Spenser would do differently
[47:00] Advice for new PMs
[50:00] Wrap-up
---
💥 To support the podcast:
1. Subscribe so you don’t miss an episode 🔔
2. Leave a great review on Apple Podcasts or Spotify ❤️
3. Join the YouTube channel

Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.

Wondering where your Product & Engineering organization stands on AI? Take our AI Maturity Check here (3 min).
Get the in-depth analysis from every episode by subscribing to the Clef de voûte newsletter.
--
In today’s episode, I’m glad to be with Marie Martens, Co-Founder of Tally.

Marie co-founded Tally, a simple yet powerful online form builder, to solve a small frustration, and ended up challenging the biggest software companies on the market without raising a single dollar of venture capital.

With over 1 million teams worldwide using their product, her bootstrapped journey proves that keeping things radically simple can be your biggest competitive advantage.

In this episode, we explore how ignoring traditional startup rules and giving away core features for free fueled massive product-led growth.

We dive into the daily realities of staying intentionally small, turning down lucrative enterprise clients, and leveraging transparent public building to generate authentic word-of-mouth.
Marie also shares her direct experience with the unexpected challenges of AI-driven customer acquisition.

We talk about :

Bootstrapping a global software product with a tiny team.
Using a viral free tier instead of hiring a sales department.
Turning down large corporate clients to maintain product focus.
Acquiring the first users through unscalable manual outreach.
Navigating the unexpected churn consequences of ChatGPT referrals.

Have a good listening !
---
[00:00] Introduction
[01:55] What's Tally ?
[02:59] Radical simplicity
[04:21] No signup friction
[06:27] ICP focus
[09:32] Saying no to enterprises
[10:09] Features not built
[12:58] Freemium model
[17:34] PLG principles
[23:58] Limits of PLG
[26:08] LLM impact
[29:19] Acquisition strategy
[31:22] Early cold outreach
[35:28] No sales team
[38:06] Brand building
[41:54] Lovable products
[46:50] Building in public
[56:42] Key hires
[59:40] AI impact
[1:05:24] Common product mistakes
---
💥 To support the podcast:
1. Subscribe so you don’t miss an episode 🔔
2. Leave a great review on Apple Podcasts or Spotify ❤️
3. Join the YouTube channel

Hébergé par Ausha. Visitez ausha.co/politique-de-confidentialite pour plus d'informations.