Case study: Deep Research feature
Paano natapos ng research team ang industry report sa loob ng ilang oras gamit ang Deep Research.
May isang business day ang four-person market research team para mag-deliver ng competitor briefing. Nag-import sila ng 12 core Sources sa NotebookLM at gumamit ng Deep Research para punan ang gaps bago mag-export ng Studio PDF.
Constraint: dapat traceable ang bawat bullet sa final deck — requirement na natural na natutugunan ng citation model ng NotebookLM.
Timeline (actual run)
| Time | Action | Notes |
|---|---|---|
| 0:00 | Mag-upload ng 8 PDF + 4 URL | Investor decks, 10-K excerpts, earnings call transcripts |
| 0:20 | Chat: scope questions | Market size, share shifts, regulatory risks |
| 0:45 | Deep Research pass | Nagdagdag ng 6 web Sources na inirekomenda ng NotebookLM |
| 1:30 | Studio Briefing Doc | Executive summary + SWOT na may citations |
| 2:00 | PDF export + review | Na-verify ng PM ang bawat citation sa loob ng 30 min |
Before vs after
| Metric | Manual process | NotebookLM Deep Research |
|---|---|---|
| Wall-clock time | 2–3 araw | ~2 oras active work |
| Citable bullets | Manual footnotes | Auto-linked Sources |
| Gap detection | Analyst memory | Deep Research suggestions |
| Stakeholder format | Rebuild sa slides | Studio export + light edits |
Lessons learned
- Magsimula sa makitid na research question — pinapalaki ng malawak na prompts ang Deep Research runtime
- Mas piliin ang structured PDFs kaysa scraped HTML kapag mahalaga ang accuracy
- Mag-second Chat pass: "ano ang sumasalungat sa thesis natin?" bago mag-export
- Magpanatili ng "source quality" note: i-flag nang maaga ang paywalled o outdated items
Team quote: "Hindi pinalitan ng Deep Research ang judgment namin — pinalitan nito ang boring assembly step para makapaglaan kami ng oras sa interpretation."