Blog
Notes on self-hosted AI, valuation workflows, and building defensible analysis for private markets.
-
A Bigger Model Is a Better Model — If You Understand Every Line
Valuation orthodoxy says keep the model small, because complexity hides errors. But the real constraint was never size — it's comprehension. Move that ceiling, and the trade-off changes.
-
A "Local" AI Agent Doesn't Mean Your Data Stays Local
Many "local" AI agents still send your documents to an external LLM. Here's why that distinction matters for valuation and confidential financial data.
-
Using AI in Valuation Reports Without Undermining Your Credibility
A practical, standards-aligned guide to using AI in valuation reports—where it earns its place, where it creates exposure, and how to keep your judgment intact.