Audio companions to each preprint — paper-length, conversational walkthroughs generated with NotebookLM. For the commute, not the desk.
Each episode maps one-to-one to a paper in the research series. The format is deliberately unhurried: two AI co-hosts work through the paper's argument the way two researchers might work through it over coffee — with room for tangents, pushback, and plain-language analogies for the trickier ideas. Average length: 18–22 minutes.
Why human marketplace reputation collapses under agentic assumptions, and how the Composite Reliability Index aggregates ten orthogonal signals into one score in [0, 1].
The structural taxonomy that separates the human-assistive AI economy from the autonomous agent-to-agent economy — and why conflating them breaks almost every attempt to size the market.
Why cryptographic proofs stop being proofs the moment they leave the chain — and the three-layer architecture of attestation, validation, and quality-market incentives that replaces them.
Five axes of disagreement across fifty-one competing definitions of "agentic economy" — why the field can't yet measure itself, and what a minimum-viable shared map would need to include.