THE WORDS BEFORE THE MACHINE
Find what prospects repeat.
Count a phrase once per company, keep the source note attached, and test whether feature jargon destroys the problem signal.
Lexical evidence only. No outreach is sent. Repetition does not prove product-market fit, causality, response rate, or willingness to pay.
Awaiting CSV
No phrase mined
DISTINCT COMPANIES-
SEGMENTS-
VALID ROWS-
Load the bundled six-note sample and make a prediction.
Method: Papa Parse validates company, segment, and note fields. Compromise normalizes each note. Five-word windows are counted once per distinct company, then ranked by source count. Synonyms and semantically similar wording are not merged.
Awaiting analysis