Research report $/report
Paste a URL (competitor page, SEC/CFTC filing PDF, or an earnings video link) or raw text. We ingest it with our own reader stack — PDF redaction-recovery, MarkItDown, TLS-impersonating fetch, and yt-dlp+whisper for video — then produce a one-pass structured research digest (thesis · catalysts · risks · hard numbers · the prediction-market angle). The digest runs on a free long-context model, so the report is fast and cheap.
Ingests via empire reader (unredact / MarkItDown / curl-impersonate / whisper). Digest via NIM (free upstream).