Alembic and the Future of AI in Marketing - Ep. 263

What It Covers

Thomas Puig, founder of Alembic, explains how his company uses spiking neural networks and causal AI mathematics to transform marketing intelligence, processing billions of data rows to connect customer touchpoints and prove campaign ROI.

Key Questions Answered

• •Spiking Neural Networks for Signal Processing: Alembic runs spiking neural networks on NVIDIA GPUs using custom wetware simulators to compare different marketing modalities like Nielsen ratings versus store visits, enabling outlier detection with zero time history for short campaigns like Olympics sponsorships.
• •Private Data as Competitive Advantage: Corporate profit follows information flow, with all future alpha coming from private datasets rather than public ones. AI models trained on similar public data converge to 90% similarity, making proprietary customer data the key differentiator like BP versus Shell gasoline.
• •Causal Chain Reaction Modeling: Alembic connects marketing touchpoints across time using causal inference mathematics to track sequences like watching Olympics coverage, searching Google flights, clicking ads, then purchasing tickets, calculating optimal lag times between each conversion step to prove campaign effectiveness mathematically.
• •LLMs for User Experience Only: Use large language models like Mad Libs templates where AI writes connecting language but causal deep learning models fill in actual data points, eliminating hallucination risks while maintaining readable intelligence briefings instead of traditional dashboards that require human interpretation.