Ever wondered what happens when a financial controller decides to stop reading AI newsletters and actually starts building?

​In this inaugural episode of the Deep Dive, our AI hosts unpack a fascinating community-submitted knowledge document detailing the creation of CosmoTender—a live, AI-powered tender monitoring system built from scratch alongside a demanding full-time corporate finance career.

​We peel back the hood on the creator's "Agentic OS" architecture, breaking down the messy, real-world engineering challenges of stitching together n8n for orchestration, Supabase for database management, and cutting-edge LLMs into a stable production pipeline that automatically monitors dense Dutch government procurement data (TenderNed). This isn't theoretical tech-bro hype; it’s a practical masterclass in automated data extraction, vector storage, and multi-agent workflows.

​But it’s not all code and databases. We also explore the human side of the shift: what it feels like to transition from a structured finance mindset to an iterative AI builder.

​To wrap up, we tackle the bigger picture: what does the rise of agentic AI actually mean for traditional office workers? We cut through the corporate marketing fluff to talk about the reality of what changes, what stays the same, and why finance and operations professionals can start building their own tools today rather than watching from the sidelines.