A 10-K is hundreds of pages and maybe five that move a thesis. Asking AI to summarize it is the one request that buries those five.
A summary smooths everything into one tidy paragraph. But a 10-K is built to look like last year on purpose. The signal lives in the diff, and a summary averages the diff away.
So before you read a page, give AI a sorting job instead:
1. What CHANGED: every item that differs from last year, with the page cited. 2. What MATTERS: of those changes, the ones that touch revenue, margins, debt, or guidance. You rank them. 3. What needs VERIFICATION: every figure gets a source, a page, and a date, or it goes on a follow-up list before you trust it.
Notice what AI is not doing: it is not deciding if the changes are good, what they are worth, or what you should do. That stays with you.
AI sorts the pages. You own the thesis.
Save this one and run it on your next filing.
Educational content only. Not investment advice, and not a recommendation to buy, sell, or hold any security. Wall Street Prompt. Always verify against the primary source filing.