WhyHow.AI

WhyHow.AI builds litigation intelligence software that helps plaintiff firms identify emerging class-action and mass-tort opportunities. Based in San Francisco, the company combines Exa’s semantic web search with LLM reasoning agents trained on legal strategies. Its platform looks for early signals of potential cases in articles, user forums and regulatory filings, connecting scattered evidence of drug reactions, automotive defects and data-privacy abuses.

Co-founders Chia Jeng Yang and Tom Smoker have also worked on tools for turning unstructured enterprise data into structured knowledge. Its WhyHow SDK, documented in 2024, let users create, manage and query graphs directly from PDFs. Users could define a schema specifying the entities and relationships to extract, giving them control over how documents became queryable data.

In joint engineering work with Neo4j, WhyHow explored combining graph traversal with vector search for financial-report analysis. The approach uses explicit relationships to guide retrieval while retaining access to the underlying text. Its chunk-linking feature connects graph nodes to corresponding document passages, allowing an LLM to receive both structured relationships and source context.