Missing startup postmortems — searchable public archive with structured data

Founders, early-stage investors, and solo operators lack a centralized, searchable repository of candid startup postmortems; learning is fragmented across blogs, tweets, and gated newsletters, which makes it hard to surface repeatable signals from failure and to answer questions like “how often do marketplaces fail after raising Series A?” There are roughly 20 million active startup teams globally and many pay $600/year for research and intelligence, so the absence of structured failure data imposes a real cost on decision-making and learning. You could build a public archive that ingests narrative postmortems and applies LLM-based extraction to produce structured facts, timelines, risk tags, and quantified lessons, coupled with full-text search, faceted browsing, and an API for investors and tooling for founders. Priced as a mix of a free searchable layer and paid subscriptions or enterprise licenses, the addressable market aligns with the $12.0B annual spend implied by 20M teams × $600/year, and the current maturity of AI summarization makes automated, scalable tagging and signal extraction feasible for the first time. This will stand out by combining rigorous curation and provenance (verified submissions, redaction/anonymization workflows), a consistent taxonomy of failure modes, and data exports/APIs that let buyers run empirical analyses—advantages over scattered blogs or closed communities. The challenges are real: sourcing high-quality, unbiased postmortems at scale, navigating legal/privacy concerns, and building trust in AI-extracted facts—but if solved thoughtfully this idea could deliver measurable decision-making value to founders and investors.