markdropDocs
Open Markdrop
All releases
ImprovedAI

Ask and search read whole transcripts, Office files and video captions

Ask and search now read every minute of a meeting's transcript, the text of Word, Excel, CSV and PDF files, and a video's captions, and answers say where each thing is: "at 1:10:00", "page 3", "sheet Budget", with links that open right there.

  • Web app
  • API
  • MCP server
  • Whole transcripts. A long meeting was searchable only for its first 20 minutes or so. Now every minute is, up to about three and a half hours per meeting. A search result shows the moment the words were said, and Ask quotes that line with a link to it.
  • Office files and PDFs. Word documents (.docx) by section, Excel workbooks (.xlsx) and CSV files by sheet, and PDFs page by page are read in the background after upload. Results and answers say where: Page 3 ·, Sheet Budget ·, and a PDF opens at that page. Password-protected Office files and encrypted PDFs are never read.
  • Video captions. Add a .vtt or .srt file from a video's ⋯ menu (Add captions). The player shows them as subtitles (CC, or press C), and search and Ask read the video minute by minute: an answer's source plays the video from that moment. A video is also found by the text around it on pages that embed it, for people who can open those pages.
  • MCP and API. set_captions adds, replaces or removes a video's captions; PUT /api/v1/artifacts/{slug}/captions does the same.
  • Existing items are brought up to date in the background, a few files at a time.