Roundtable: Building Reliable AI Agents | Dust, Rippletide, Hugging Face, WorldGame
🤖The AI agent revolution is here. But most are building toys, not tools.
Join us for an exclusive Technical Roundtable on Building Reliable AI Agents, hosted by 42 Artificial Intelligence and powered by Rippletide.
✨ EXPERT PANEL
While everyone else demos chatbots, these four experts are building systems that actually work:
• Stanislas Polu - Co-founder, Dust (ex-OpenAI, ex-Stripe)
• Yann Bilien - Co-founder & CSO, Rippletide (Télécom Paris, Imperial College London)
• Hugo Hernandez - Co-founder, WorldGame (ENS Ulm, HEC Paris)
• Amir Mahla - ML Research Engineer, Hugging Face (ex-H Company)
🔧 WHAT WE'LL EXPLORE
Technical architecture decisions, reliability patterns, and production challenges that separate functional AI agents from research prototypes. From multi-agent orchestration to handling edge cases at scale.
🎯 WHO SHOULD ATTEND
Engineers, researchers, and ML practitioners working on agent systems, plus students passionate about AI who want to learn from industry experts.
FAIR WARNING This is a technical roundtable. If you're still figuring out why your agents hallucinate in production, you're in the right place. If you think "agent reliability" means restarting when it crashes, this might be too advanced 🤖
LIMITED SPOTS - APPROVAL REQUIRED Register now to try to secure your spot!
📅 EVENT AGENDA
• 6:45-7:15 PM - Doors Open
• 7:15-7:20 PM - Welcome and speaker introductions
• 7:20-8:05 PM - Expert panel discussion
• 8:05-8:20 PM - Technical Q&A session
• 8:20-8:25 PM - Rippletide Demo
• 8:25-8:30 PM - Closing remarks
• 8:30-11:00 PM - Afterwork - Innovation meets disruption over drinks (Salle Damogran with terrace)
MEET OUR SPEAKERS
🚀 Stanislas Polu • Co-founder, Dust
Co-founded Dust after selling his first company to Stripe and spending three years at OpenAI advancing LLM reasoning. While others theorize about AI agents, Stan deploys them at enterprise scale. Companies like Clay, Doctolib, and Alan use Dust's AI agents to achieve 70% weekly engagement (matching Slack), save 50,000+ hours yearly, and scale teams 4x. Stan is an alum of Polytechnique and Stanford.
🧠 Yann Bilien • Co-founder & Chief Scientist, Rippletide
As Co-Founder and Chief Scientist of Rippletide, Yann Bilien leads the development of a Hypergraph Database for AI agents delivering 99.9% accuracy, hallucination-free and full explainability for enterprises.
After graduating in Mathematics from Télécom Paris, he conducted research at Imperial College London on proximal methods for deep neural network training. Yann pioneered the development of AI agents based on world models and continues to pursue cutting-edge, cross-disciplinary technologies.
🎮 Hugo Hernandez • Co-founder, WorldGame
Co-founder of WorldGame, where he held the role of CSO. He single-handedly built Koine, an AI agent that could prototype games and websites from natural language and reproduce entire design systems. At WorldGame, he led Projet Thanos (Tool for Human Annotation and Normalized Observation Sequences), which won the prestigious I-Nov competition from BPI with €850K in funding, and he guided the development of GAMA, a multi-agent framework that improves AI system reliability from 55-61% to 95% accuracy. Hugo studied at ENS Ulm (École normale supérieure) and HEC Paris.
🤗 Amir Mahla • ML Research Engineer, Hugging Face
Machine Learning Research Engineer at Hugging Face, building full-stack AI agent systems including smolagents framework, screensuite evaluation, and open-r1. Previously at H Company, he built next-generation agentic models for enterprise automation and co-authored the Holo1 paper. After transitioning from professional music to AI through École 42 Paris, Amir specializes in end-to-end agent solutions from pre-training LLMs/VLMs to production deployment, solving the hard integration problems that make AI agents work in complex environments.
About Rippletide
Rippletide has cracked the code on AI agent reliability. While others deliver unpredictable chatbots, Rippletide's breakthrough hypergraph decision engine achieves 99% accuracy with zero hallucinations through deterministic reasoning layers - solving the fundamental reliability problem that prevents enterprise AI deployment.
The results: pilot customers report +38% more meetings and sub-1-minute speed-to-lead, with AI agents that stay on script, stay compliant, and stay alive in production with line-by-line rationale in real time.
Founded by serial entrepreneur Patrick Joubert (three exits: Recast.AI → SAP, Ponicode → CircleCI) and Yann Bilien, neuro-symbolic AI researcher from Imperial College London.
Rippletide offers an enterprise-grade Hypergraph database for AI agents that is auditable, explainable, and GDPR-compliant from day one.
About 42 Artificial Intelligence
42AI is École 42's AI hub, developing AI expertise through technical bootcamps, research labs, and industry partnerships.
École 42 is France's most selective programming school and part of the global 42Network, producing top-tier technical talent through its innovative peer-to-peer model. Founded by Xavier Niel, whose investment portfolio includes key players like Mistral, Kyutai, and H Company, École 42 sits within France's AI innovation landscape.
Through Machine Learning and Deep Learning bootcamps, hands-on lab projects, and strategic industry collaborations, 42AI transforms École 42's exceptional programming foundation into AI expertise to drive France's and Europe's growing AI ecosystem.