Subscribe →
Video to Text 2 Jun 2026

How to Build a Searchable Brain from Your Zoom Meeting Archives

Time saved: Hours of re-watching old recordings. Value unlocked: Every insight, decision, and idea you’ve ever discussed — findable in seconds.

You have 200 Zoom recordings sitting in the cloud. Somewhere in there is the client who told you exactly why they bought. The strategy call where you landed on the idea that actually worked. The feedback session where someone said something that should have changed everything.

You’ll never find it — because searching video is basically impossible.

This workflow fixes that. By the end, you’ll have a fully searchable knowledge base built from your entire meeting archive: every transcript, indexed and queryable through plain English. No coding. No expensive software. Just a smart combination of tools you probably already have access to.

The Stack: What You Need

  • Zoom Cloud Recordings (or local recordings saved to Google Drive/Dropbox)
  • Fireflies.ai or Otter.ai — for bulk transcript generation (Fireflies preferred for this use case)
  • Notion — free plan works — as your knowledge base
  • Claude — for synthesis and querying
  • Optional: Zapier or Make to automate future recordings into your database automatically

Step 1: Export Your Existing Zoom Transcripts

If You Have Zoom Cloud Recording with Auto-Transcription Enabled

  1. Log into your Zoom web portal and go to Recordings in the left sidebar.
  2. Click into any recorded meeting. Under the recording, you’ll see a “Transcript” file (.vtt format).
  3. Download the transcript. Repeat for each recording you want to include.

Heads up: Zoom’s native transcription is decent but not the cleanest. If you have a backlog of 20+ meetings, consider uploading the recordings directly to Fireflies instead for better transcript quality.

If You’re Using Fireflies for Past Recordings

  1. In Fireflies, go to Uploads in the left menu.
  2. Click “Upload Audio/Video.”
  3. Drop in your Zoom .mp4 files — Fireflies will transcribe them automatically, usually within 10–20 minutes per hour of audio.
  4. Once processed, go to each meeting and click “Transcript” → “Export” → choose Plain Text or SRT.

If Your Recordings Are on Google Drive

  1. In Fireflies, connect your Google Drive under Settings → Integrations.
  2. Fireflies will scan for audio/video files and offer to transcribe them in bulk.
  3. Approve the batch and let it run.

Step 2: Standardise Your Transcripts Into a Consistent Format

Raw transcripts are messy — timestamps, speaker tags, filler words. Before building your knowledge base, clean them up using Claude.

For each transcript (or a batch of smaller ones), run this prompt:

Here is a raw meeting transcript:

[PASTE TRANSCRIPT]

Please reformat it as follows:
1. Remove all timestamps and timecodes.
2. Keep speaker names but clean up obvious filler words (um, uh, like, you know).
3. Break the content into labeled sections based on topic shifts. Use H3 headers for each section (e.g., ### Project Status Update, ### Budget Discussion).
4. At the top, add a metadata block in this format:
   - Date: [extract from transcript if present, or write "Unknown"]
   - Attendees: [list all speakers]
   - Meeting Type: [e.g., Client Call, Team Standup, Strategy Session — infer from context]
   - One-Line Summary: [write a single sentence describing the meeting's main purpose and outcome]
5. At the bottom, add:
   - Key Decisions Made: [bullet list]
   - Action Items: [bullet list with owner names]
   - Topics to Follow Up On: [bullet list]

Return only the reformatted transcript.

This turns a raw wall of text into a structured document you can actually use.

Step 3: Build Your Notion Knowledge Base

Create the Database Structure

  1. In Notion, create a new page called “Meeting Brain” or “Call Archive.”
  2. Add a new Database (Full Page) — choose Table view.
  3. Set up these columns:
    • Name (default) — Meeting title
    • Date — Date property
    • Meeting Type — Select property (Client Call, Strategy, Team Standup, Sales, etc.)
    • Attendees — Text property
    • One-Line Summary — Text property
    • Tags — Multi-select (add tags like: pricing, objections, product feedback, roadmap, etc.)
    • Full Transcript — Each row opens as a page — paste your cleaned transcript here

Populate the Database

  1. For each cleaned transcript from Step 2, create a new row in your Notion table.
  2. Fill in the metadata columns (Date, Type, Attendees, Summary) — copy these directly from the metadata block Claude generated.
  3. Open the row as a full page and paste the complete cleaned transcript inside.
  4. Add relevant Tags — these become your filtering superpower later.

If you have 50+ meetings to import: Don’t do this manually. Jump to Step 5 (automation) and use Zapier to auto-create Notion rows from new Fireflies summaries going forward. For the backlog, batch 10 transcripts at a time.

Step 4: Query Your Knowledge Base with Claude

Here’s where it gets powerful. Once your transcripts are in Notion, you can copy batches of them into Claude and interrogate your entire call history.

The Search Query Prompt

I'm going to paste several meeting transcripts from my archive below. After reading all of them, please answer the following question:

QUESTION: [e.g., "What were the most common objections clients raised about pricing?"]

Instructions:
- Only use information from the transcripts I provide. Do not add outside knowledge.
- For each relevant finding, cite which meeting it came from (use the meeting date and attendees from the metadata block).
- Format your answer as: Finding → Source Meeting
- If a theme appears in multiple meetings, group them together.
- End with a synthesis paragraph summarising the overall pattern you found.

TRANSCRIPTS:
[PASTE 3–5 TRANSCRIPTS HERE]

Example Queries That Work Well

  • “What product features have clients asked for most often?”
  • “Which objections keep coming up on sales calls?”
  • “What did [client name] say about their main pain points across all our calls?”
  • “Summarise every time we discussed pricing strategy.”
  • “What commitments did I make to clients that I haven’t followed up on?”

Step 5: Automate Future Meetings Into Your Brain

Once your archive is built, set it up so every new meeting automatically flows in.

Zapier Automation (Recommended)

  1. Go to Zapier and create a new Zap.
  2. Trigger: Fireflies — “New Meeting Transcript Ready”
  3. Action 1: Claude (via Zapier’s Claude integration) — run the cleaning and formatting prompt from Step 2 automatically.
  4. Action 2: Notion — “Create Database Item” — map the output fields (summary, date, attendees, transcript) to your Meeting Brain columns.

Result: Every meeting you have from now on appears in your knowledge base within 30 minutes of the call ending — with no manual work.

The Master Prompt: Monthly Intelligence Digest

At the end of each month, copy your last 30 days of meeting transcripts into Claude and run this:

Below are all my meeting transcripts from the past month. Please generate a Monthly Meeting Intelligence Report with the following sections:

1. DECISIONS LOG: Every decision made across all meetings, grouped by project or client.
2. OPEN LOOPS: Any action items, questions, or commitments that were mentioned but I haven't seen a resolution for.
3. RECURRING THEMES: Topics, concerns, or ideas that came up in 3 or more meetings.
4. CLIENT SIGNALS: Any strong feedback — positive or negative — that clients expressed.
5. STRATEGIC INSIGHTS: 3 things I should be paying attention to based on patterns in these conversations.

Be specific. Reference dates and attendees when citing examples.

TRANSCRIPTS:
[PASTE ALL MONTHLY TRANSCRIPTS]

Expected Output: What “Done” Looks Like

When this system is fully set up, you’ll have:

  • A Notion database with every meeting from your archive, cleanly formatted and tagged
  • Instant filterability — click a tag like “pricing” and see every meeting where it came up
  • A Claude-powered query layer — paste in transcripts and ask anything about your call history
  • A monthly digest that surfaces patterns you’d never catch meeting by meeting
  • Automatic ingestion of new meetings going forward (if you set up the Zapier flow)

The first time you search your archive and surface a specific insight from a call 8 months ago — one you completely forgot — you’ll understand why this is worth the 2 hours it takes to build.