Lenny RAG / EigenPM Explorer

A two-headed project: a retrieval-augmented Q&A system over Lenny’s Newsletter content, and EigenPM Explorer — a 3D interactive map of 297 product managers from Lenny’s Podcast, projected into psychographic space via PCA.

The interesting bit is the second one — and the methodology choice inside it. I extracted entities and relationships from each interview (a GraphRAG-style pipeline), then built the guest space two ways: from what people talk about (entities) and from how they connect ideas (relationships). The relationship space explained meaningfully more of the variance — roughly 26% vs. 15% in the top components — so that’s the space the map lives in. PCA reduced 1,500+ dimensions to the top 20 along which Lenny’s guests naturally vary, rendered as a Three.js scene you can rotate and click through; a short quiz places you on the map alongside the guests you cluster with.

The honest caveat: the retrieval half shipped without a formal eval suite — building one (retrieval quality and groundedness, measured rather than assumed) is the project’s open to-do, and the write-up will land here when it does.