Chapter 7 — Workflow: Fathom Manual Lookup

Category: Workflows Reading time: 5 minutes

---

Purpose

Two utility workflows for manually fetching transcripts from the Fathom API. Used for testing, backfilling historical meetings, or debugging. These workflows are never activated — they run only when manually triggered from the n8n UI.

Workflow A: List Recordings

Workflow file: workflows/fathom-list-recordings.json n8n Workflow ID: 3

Node Flow

Manual Trigger
    → Code: Set Date Parameter
    → HTTP Request: List Meetings by Date
    → Code: Format Recording List

Usage

1. Open the workflow in n8n
2. Click the “Code: Set Date Parameter” node
3. Change targetDate to your desired date (e.g., '2025-09-02')
4. Click “Test Workflow”
5. Check the output of the last node for a list of recordings

Output Format

Each recording shows:

Field	Description
recording_id	Unique ID needed for Workflow B
title	Meeting title from calendar
start_time	When the recording started
duration_minutes	Computed from start/end times
url	Shareable Fathom link

Workflow B: Fetch Transcript

Workflow file: workflows/fathom-fetch-transcript.json n8n Workflow ID: 4

Node Flow

Manual Trigger
    → Code: Set Recording ID
    → HTTP Request: Get Meeting Details
    → Code: Find Meeting and Extract Date
    → HTTP Request: Fetch Transcript
    → Code: Format and Save Transcript

Usage

1. Run Workflow A to get the recording_id
2. Open this workflow in n8n
3. Click “Code: Set Recording ID”
4. Set recordingId to the value from Workflow A
5. Set targetDate to the same date (needed to filter the meetings API)
6. Click “Test Workflow”
7. Check the output — it tells you whether the matching chat log was found

Why Two Separate Workflows?

n8n doesn’t natively support interactive “pick from a list” mid-workflow. The two-step approach is simple: list recordings, note the ID, fetch the transcript. For occasional testing use, this is more reliable than a complex single workflow.

Design Note: Sequential Over Parallel

The original design used parallel HTTP requests (transcript fetch + meeting details simultaneously) merged via an n8n Merge node. This failed in n8n 2.10.4 — the Merge node’s mergeByPosition mode requires match fields that don’t apply to this use case. The fix was to make the flow purely sequential, using $('NodeName').item.json references to access data from earlier nodes. Simpler and more reliable.

Backfilling Historical Meetings

To process an old meeting:

1. Run List Recordings with the meeting date
2. Run Fetch Transcript with the recording ID
3. Drop the matching chat log into ~/n8n/watch/YYYY-MM-DD-zoom-chat.txt
4. The Merged Call Summarizer’s Local File Trigger will detect the new file, find the transcript, and run the full pipeline