DataGen

DataGen is a French-language interview podcast about how companies build with data and AI, hosted by Robin Conquet. The format almost never varies. One guest, one set of problems, one conversation that keeps going until the mechanics are clear. Conquet opens by laying out who the guest is and what they have shipped, then hands the floor over. Guests are practitioners rather than commentators. A Head of Machine Learning at Gorgias walks through how the arrival of generative AI pushed his team to rebuild the LLM architecture and put those features at the centre of a customer support platform serving e-commerce merchants. A VP Data at Accor describes changing his team's operational model and standing up a factory that ships data and AI solutions to the business. Axel de Goursac, a partner AI lead at KPMG France who previously directed data science and AI at LVMH, works through what a generative AI strategy needs in order to scale and produce measurable return. The technical passages are the point, not decoration. One guest details two parallel pipelines, each generating its own conclusion, then merged into a single prompt so an agent can decide whether to close a conversation or open it further. Another explains why the modern data stack matters when analytics is exposed to customers inside a product rather than to internal dashboards. A third describes American back-end engineers casually contributing to dbt projects, and argues the AI engineer sits on the border between data work and back-end work. Episodes tagged as masterclass narrow the aperture to one discipline. Matthieu Rousseau, who founded an agency staffed entirely by data engineers, returns to the modern data stack. Yoann Benoit, co-founder at Hymaia, makes the case that scaling AI agent adoption is a leadership problem first, and that safe spaces for experimentation collapse when a manager preaches transformation without ever trying the tools. That mix of pipeline detail and organisational reality is what gives the show its range. Christophe Bleffari, who writes a well-known French data newsletter and spent three months at Y Combinator after co-founding Nao, compares how data teams operate in the United States against how they operate in France. Conquet closes every episode the same way. He asks each guest what their biggest difficulties are and which resources they recommend, then centralises the answers into files distributed through the DataGen newsletter, alongside a short Thursday email summarising the key points of the episode. Some instalments carry a sponsor. The register stays collegial and unhurried, with Conquet often noting that he has worked with a guest before. For anyone running or joining a data function in France, it is a running record of what the job currently involves.