Fathom records and transcribes calls. Your agent reads the transcript, pulls the summary and the action items, and writes them onto the candidate in your CRM. The debrief you were going to type up is already written.
Fathom is a read source. The transcript comes in; structured notes go out to the CRM.
things you can ask for
- “Write up this morning's interviews into Recruit CRM”
- “What did the client say they actually wanted? Turn it into a brief”
- “Take the brief from this call and source against it with Apollo”
- “Pull the action items from that call”
works well with Fathom
- Recruit CRM
the debrief written onto the candidate
- Apollo
source directly from what the client said on the call
- Vincere
interview notes onto the record
- Granola
the other notetaker, if your team runs both
Skills that use Fathom
A skill is a whole pipeline your agent already knows. These ones call Fathom as part of the run.
Reverse poach → original client alert
Placed candidate updates LinkedIn. Track via AI Ark, log Fathom call notes, alert the original client.
What your agent reads before it touches Fathom
Every tool ships with a written playbook, and the agent loads it before the first call. Auth, rate limits, what each call costs, which actions need your approval, and the mistakes worth avoiding. It is the difference between an agent that knows the tool and one that guesses at it.
Read the Fathom docs →Show the raw playbookwritten for the agent
HOW in lib/fathom.py. Docs: developers.fathom.ai/quickstart.
Auth & config
- Base URL:
https://api.fathom.ai/external/v1· Auth: headerX-Api-Key(envFATHOM_API_KEY; never hardcode). - User-scoped: a key only sees meetings recorded by / shared with that user.
- Pagination: cursor-based —
next_cursorin the response → pass back ascursor.iter_meetings()auto-walks it.
Operations (read-only API)
list_meetings (filters: created_after/created_before ISO 8601, recorded_by emails, include_transcript), iter_meetings (auto-paginate), get_transcript(recording_id). Meeting objects carry title, URL, timestamps, summary, action items, CRM matches.
Handoff
Read source: post-interview transcript/summary → parse → write notes into the ATS (Recruit CRM / Bullhorn / Loxo …) as a candidate activity. Pairs with [[granola]] (also a notetaker) — both are API-accessible; Granola needs a Business plan.
Callable surface — lib/fathom.py
Import: from lib.fathom import Fathom → instantiate Fathom() (reads key from env). Base: https://api.fathom.ai/external/v1. Generic passthrough: request(method, path, *, params, json).
get_transcript(recording_id: str) -> Any— GET /recordings/{recording_id}/transcript — fetch a transcript separately.iter_meetings(**params) -> Iterator[dict]— Auto-paginate meetings via next_cursor.list_meetings(*, include_transcript: bool | None = None, recorded_by: list[str] | None = None, created_after: str | None = None, created_before: str | None = None, cursor: str | None = None, **params) -> Any— GET /meetings — filter by date (ISO 8601) / recorder email; optionally inline transcripts.