About
Former diplomat, with years of experience navigating crises (sometimes war, literally)…
Articles by Romain
-
Silicon Valley Etiquette
Silicon Valley Etiquette
Manners Matter. “During the 17th century, in France, manners became a political issue.
141 Comments
Activity
5K followers
-
Romain Serman reposted thisRomain Serman reposted thisAI agents have entered the workforce. But can they actually do the job? Introducing APEX-Agents, our new benchmark that evaluates whether agents can complete real, long-horizon professional services deliverables in Google Workspace. We surveyed hundreds of professionals from McKinsey, Goldman Sachs, and Cravath to understand how they spend their time. Then, we worked with them to create 33 scenarios and 480 tasks across investment banking, consulting, and corporate law. This work reflects the deliverables that a professional would spend hours or days completing. We partnered with Box to simulate realistic enterprise file systems by mapping rigorous, domain-specific challenges to the complex structures professionals use every day. We also worked with Harvey to validate that the legal tasks reflect the work of top lawyers at Fortune 500 enterprises and law firms. Then we tested today’s best models. Mercor APEX-Agents Pass@1 leaderboard: 🥇 Google DeepMind Gemini 3 Flash (High): 24.0% 🥈 OpenAI GPT-5.2 (High): 23.0% 🥉 Anthropic Claude Opus 4.5 (High): 18.4%
-
Romain Serman reposted thisRomain Serman reposted thisAI being smart isn’t the same as it being useful for real work. AI can pass the bar exam. But can it redline a contract? That’s one of questions we set out to answer with some of the world’s top industry experts: - Lawrence H. Summers, former US Treasury Secretary - Cass Sunstein, the most cited legal scholar - Eric Topol, MD, physician and best-selling author - Dominic Barton, former global managing partner of McKinsey - Dozens of professionals from Goldman Sachs, McKinsey, Latham & Watkins, Mount Sinai, and more Today, we're launching the Mercor AI Productivity Index (APEX), a benchmark that measures how well AI has automated the most valuable industries in the world. Most benchmarks measure abstract capabilities. APEX evaluates model performance on real deliverables—the same work consultants, bankers, lawyers, and doctors produce every day. Key findings from APEX: - GPT 5 demonstrates the strongest performance across all 4 domains - Some cheaper models outperform more expensive models from the same provider (e.g. Gemini 2.5 Flash vs. Gemini 2.5 Pro) - The best open source model, Qwen (7th), performs only 2% behind Grok 4 overall This is just the beginning. With APEX v2, we’ll expand to more industries, more complex tasks, and provide an even sharper picture of how AI is reshaping the economy.
-
Romain Serman reposted thisRomain Serman reposted thisBeing a founder for 6 months, there’s one thing I didn’t expect. How many people are willing to help you with nothing in return. Six months ago, I knew almost no one in sales. We had just decided to build Caretta, so I started reaching out to VPs of Sales for validation. My message was simple: - I’m a young founder - Building a product around this pain point - I have a few questions; could we hop on a call? Most didn’t reply. But some did. And those “some” turned out to be seriously successful leaders—running teams up to 100+, bringing in millions a year. Yet they still made the time to talk to me. Over time, I realized: people genuinely like helping, with nothing in return. Good karma. When I was in Silicon Valley, I came across an article by Romain Serman that said the same thing: good karma is a currency. It stuck with me. Fast forward—some of those calls turned into real relationships. Now: - A strategist at one of the biggest UK companies mentors me every two weeks on challenges in my startup journey - Sales leaders pick up, challenge me, and help push our product forward And here’s my conclusion: At the end of the day, it’s all about human relations. Help others without expecting anything. And genuinely appreciate those who do the same for you. PS. I strongly suggest you read that article, even if you're not in SF (yet): Link in the first comment.
-
Romain Serman posted thisNot so long ago, we saw AI as our “co‑pilot,” meaning AI would fix the errors we humans make. Actually, we got it backwards: we are the AI’s co‑pilot. Our job is to fix the AI’s mistakes. That's pretty good news (for now). We still have jobs!
-
Romain Serman shared thisYou want to make it in America? Then you have to decode America. Carlotta nailed it, a must-read 👇🏻 Thanks, Lotti!Romain Serman shared thisAmerica is the land of IYKYK. An immigrant founder recently asked me: “Do you have to be from the system to succeed in Silicon Valley?” Short answer: no. But you do need to decode it. You can be brilliant and still have zero idea how the U.S. works. From immigration to frat houses, credit scores to college admissions, this country runs on invisible codes. That context is rocket fuel. Especially in tech and VC. Americans start the race with the playbook. Language fluency + a few connections = a head start. (Yes, they still hustle hard.) Immigrants? We show up with no script, thick accents, and a countdown clock before our visas expire. When I landed from Italy 20 yrs ago, I didn’t know you could apply to more than one college. Back home you’re tracked into a single path at 14. Options? What options? School, social life, careers... all black boxes. See the pic of my freshman-year for my (failed) attempt at cracking it. My culture shock included: ❓My first Costco run (bigger really is a mindset) ❓What is greek life ❓How to approach professors after class ❓Hunting rec letters for jobs & grad school ❓Learning that confident self-promotion is an American art form I adapted by surrounding myself with other immigrants who’d done it ( 🙏 Olivier, Santi, Fede) and generous locals who US-splained things when it counted (🙏 Courtney, Jonathan, Ya, Nicholas, Misha, Amy). I arrived thinking raw hard work was enough. It isn’t. Skill and grind are necessary but not sufficient. What matters just as much is knowing who holds the information. Now I'm deep in the system. How? Instead of generic advice, here’s my blooper reel: 👉 I was too shy to ask for help. I watched classmates build mini “boards of directors” while I brute forced assignments. Lesson: Most people love being asked, esp if you come prepared and follow up. 👉 I judged American self-promotion. The let-me-tell-you-how-great-I-am mindset felt 🤮 cringey next to European understatement. Lesson: Visibility isn’t vanity; it’s oxygen. Opportunity can’t find you if no one knows what you’ve done. 👉 Slipped into victim mode. Legacy admissions and uncle-runs-the-fund tales made the game feel rigged. Lesson: Outsider perspective is a feature not a bug. Once you decode the rules, you can remix them... and that’s where original thinking (and venture returns) are born. If any of this sounds familiar, welcome to the club! Today, as a partner at Emergence Capital, it’s my turn to pull others up the mountain. And I love it. Few things beat watching talent thrive when the odds say they shouldn’t. Immigrants make America work. If you’re on the journey, I see you. And if you’re building a B2B AI or software decacorn, DM me. Let’s bend the system together.
-
Romain Serman reposted thisRomain Serman reposted this"Mirage PMF" is one of the most common startup killers. Here's what it looks like: You hit $1–2M ARR. Growth is fast, VCs are lining up with term sheets, and you double down—scaling pipeline gen and ramping up AE hiring. But soon, the rocketship starts to rattle. More often than not, it explodes. The root cause is usually a failure to focus—an undefined Ideal Customer Profile (ICP). In the rush to grow, you end up with a hodgepodge of customers, each using your product differently. It feels like optionality, but really, it leads to a scattered product roadmap, unfocused GTM, and an overwhelmed CS team. ⏳ Building a successful startup is like navigating an hourglass. Early on, you experiment broadly, testing different customer profiles to find where your product solves the most acute pain. You can reach $1–2M ARR serving a mix of personas. But once you hit this milestone, the game changes. Now, it’s time to analyze your customer data relentlessly: Where is your PMF strongest? What specific company traits and buyer personas give you 75%+ confidence you’ll both close the deal and truly delight the user? This is the moment to narrow your focus—and it feels hard. Saying no to potential revenue is never easy. But alignment around a clear ICP is what enables you to scale efficiently. Product, marketing, sales, and CS all row in the same direction. Only then do you earn the right to methodically expand to new ICPs, growing your TAM in a disciplined way. “Mirage PMF” has always been a risk for early-stage startups. But it’s getting worse in the GenAI era, where hitting $1–2M ARR is faster than ever. I expect even more post-Series A implosions as a result. -- I'm thinking of writing a series on the most common reasons startups fail after Series A. My goal: help founders avoid these traps (even if it’s a bit of a cautionary tale). Let me know what challenges you want me to cover—and if you’d find this series useful!
-
Romain Serman reposted thisRomain Serman reposted thisAfter months of mystique, Moonvalley 𝐜𝐨𝐦𝐞𝐬 𝐨𝐮𝐭 𝐨𝐟 𝐭𝐡𝐞 𝐬𝐡𝐚𝐝𝐨𝐰𝐬! Today we get to start shouting about all the amazing work done behind the scenes 💪 We’re turning 𝘪𝘮𝘢𝘨𝘪𝘯𝘢𝘵𝘪𝘰𝘯 𝘪𝘯𝘵𝘰 𝘳𝘦𝘢𝘭𝘪𝘵𝘺, teaming up with the visionary minds at Asteria Film Co to develop the world’s 𝐟𝐢𝐫𝐬𝐭 𝐀𝐈 𝐯𝐢𝐝𝐞𝐨 𝐦𝐨𝐝𝐞𝐥 𝐭𝐫𝐚𝐢𝐧𝐞𝐝 𝐞𝐱𝐜𝐥𝐮𝐬𝐢𝐯𝐞𝐥𝐲 𝐨𝐧 𝐨𝐰𝐧𝐞𝐝 𝐨𝐫 𝐥𝐢𝐜𝐞𝐧𝐬𝐞𝐝 𝐝𝐚𝐭𝐚. Meet Marey – an AI model built for 𝐚𝐫𝐭𝐢𝐬𝐭𝐬 𝐟𝐢𝐫𝐬𝐭. Designed to empower creators, putting the power directly in the hands of those who dream, direct, and design the future of storytelling. This is just the beginning. With an absolutely top-tier team, we’re shooting for the moon—and beyond! 🚀 Naeem John Mateusz Mikołaj Bryn Sarah Michelle
-
Romain Serman reposted thisRomain Serman reposted thisIt's been a transformational month for me in terms of my understanding of startups (and I've been in B2B startups for a while, as well as on the other side of the table as an investor). A few observations: 1. People completely misunderstand why founder-led sales matter. It's not because founders are inherently better sellers to get deals across the line. It's actually because they need to engage with customers in the ACTIVE BUYING PROCESS to figure out how the business and product need to unfold. The techniques for this are poorly taught to most would-be entrepreneurs and differ dramatically from a normal sales process. 2. In B2B, Discovery calls that aren't part of the sales process don't count. This is where most B2B product managers completely fail. Going off the wrong data set. The amount of time I've wasted in my career talking to whichever customers would sit down with me and translating their pain points/feature requests to product requirements is sad. It was the wrong data set (because they weren't real buyers) and didn't surface demand. 3. Pain points are often secondary in most B2B sales. Life is a series of pain points we mostly do nothing about, and definitely don't buy software for (thanks Rob Snyder for the aha moment on pain points not mattering). Pain points do not = demand. What does = demand? Next post coming shortly... #entrepreneurs #startups #vc #ai #hbs
-
Romain Serman shared thisI'm attending SuperVenture North America 2024, taking place on November 11 at The Marriott Marquis, Times Square, New York. Will I see you there? #SVNA SuperReturn
Romain Serman commented on a post
2w
Congrats!!!!!
-
Romain Serman reacted on thisRomain Serman reacted on thisNews: We’re moving NFX to self-funded. Taking it private. No more external LP’s after fund IV. The four GP’s stay intact. We have 8 more big seed investments to make in Fund IV, so hit us up. After that, we'll be doing $100K - $3M investments with our own money. Thanks to Homebrew who led the way! NFX Gigi Levy-Weiss Pete Flint Omri Amirav-Drory https://lnkd.in/gCB_n_C7
-
Romain Serman liked thisRomain Serman liked thisPrécision pratique : j’ai vérifié hier la disponibilité réelle de mon dernier livre, "Qui pense quand je pense" à Paris (pas seulement son référencement) : - Librairie de Sciences Po : rupture de stock ; réassort commandé. - La Procure, Paris 6e : en stock. - Librairie Tschann, Paris 6e : en stock. - Librairie Lamartine, Paris 16e : en stock. - Le Comptoir des mots, Paris 20e : en stock. Plusieurs villes de région ont également déjà au moins une librairie avec des exemplaires disponibles. Et, lorsqu’il n’est pas encore en rayon, il est bien sûr commandable chez la plupart des libraires (ou en ligne). Ou en version numérique sur le site de l'éditeur : https://lnkd.in/esHEyBRm
-
Romain Serman liked thisI couldn’t be more excited to work with Edward Hu. As the inventor of LoRA (fine-tuning) and a core contributor to o1 at OpenAI, few people have Edward’s combination of technical depth and vision. Our research team is shaping how human intelligence will power the AI economy, which is the most important problem of the decade. More soon 🚀
-
Romain Serman liked thisRomain Serman liked thisThe extent of lying / exaggerating among 16-24 year olds pitching VCs for their first round or two is at an all time high right now. It’s ok to be ambitious and optimistic. Not ok to lie about VC termsheets that one didn’t receive or to talk about a verbal LOI from a customer as a signed scaler contract. Yes, we VCs are often lazy, yet some of us still do a few minutes of due diligence before we make an offer to invest.
Experience & Education
Publications
-
France, The New Startup Factory
Medium
See publicationThe French tech scene is booming, with an increasing number of entrepreneurial successes in the US. What factors drive such entrepreneurial boom ?
Languages
-
French
Native or bilingual proficiency
-
English
Full professional proficiency
Recommendations received
1 person has recommended Romain
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content