Felix Rösner - Marble AI | How Marble’s AI Clones Uncovered 40% Redundant Work. And Saved 250+ Hours/Month

Felix Rösner, Co-founder of Marble AI, builds AI-driven workflow “clones” that reveal what really happens inside complex enterprise operations. His team helps private equity operators and middle-market leaders uncover hidden inefficiencies, quantify real ROI, and automate with precision instead of guesswork. In this episode, Felix shares why deep context—not tools—is the final moat in AI-enabled organizations.You’ll LearnHow capturing keystrokes and screen activity uncovers the real workflow variations that break most automation efforts.Why PE roll-ups routinely misjudge operational reality—and how AI clones surface true time, cost, and redundancy patterns.How sharing internal best practices (revealed through data) can unlock major efficiency gains even before automation begins.How to translate thousands of workflow observations into programmatic specifications for reliable, end-to-end digital workers.The specific types of repetitive, customer-facing tasks where automation consistently returns 100–250 hours per month to teams.Chapters00:00 Intro01:03 Felix’s background and Marble AI’s mission03:12 Why context—not tools—is the last AI moat05:06 What it means to instrument real workflows07:30 The insight that sparked Marble’s approach10:31 Why PE roll-ups struggle to see operational reality12:24 AI clones vs. traditional process diagrams13:34 Balancing workflow capture with privacy and compliance15:21 Signals a company doesn’t understand its operations17:03 Case study: diagnosing a PE-backed roll-up20:40 Deploying the capture stack across teams23:43 What Marble does with the collected data26:27 Findings: redundancy, best practices, and ROI34:05 The 40% redundancy insight37:26 Practical advice for operations leadersLinks• Watch on YouTube: https://youtu.be/sc-A0MPiZxE• Connect with Felix Rösner: https://www.linkedin.com/in/felix-roesner/• Explore more interviews: https://www.justcurious.io/articles/how-marbles-ai-clones-uncovered-40-redundant-work-and-saved-250-hours-monthProof of Work is the podcast of Pluris, a platform connecting investors and operators with the world's leading applied AI experts. Each episode features the builders, operators, and investors who've actually put AI into production, turning it from buzzword to bottom line through sharp case studies and practical conversations. We explore how AI is used to grow revenue, expand margins, improve operations, and create measurable value inside real businesses.Learn more at checkpluris.comExplore more interviews and connect with experts at https://www.checkpluris.com Subscribe to our newsletter at https://proof-of-work-ai.beehiiv.com/