How Willow refuses a verdict.

brief.finance is a stock-research workspace with an enforced research discipline around its AI. Willow is the research agent; the brief is the artifact. This page is the operating spec — what the workspace promises to do on every brief, what it refuses to do, and how the system enforces both.

What we don't do

Every brief is educational research. We never give a directional verdict on a security, never publish a numerical forecast or rating, and never tell you what to do with your money.

Both sides, by construction

Every brief has a bull case AND a bear case, every time. The structure is enforced server-side; the model can't publish a brief without both.

Source pipeline

SEC EDGAR filings, earnings transcripts, TTM fundamentals, daily price action, recent news. Sources are fetched fresh on every fire — every paragraph in a brief leans on a public source we name and link to.

Forbidden-vocabulary validator

Before any brief reaches your screen, it passes through a validator that scans for directional language. Hits fail the brief and the pipeline regenerates.