n8n-nodes-plaud-cloud

This is an n8n community node. It lets you use the Plaud AI voice recorder cloud in your n8n workflows.

Plaud sells pocketable AI voice recorders that automatically sync recordings to a cloud where they are transcribed and summarized. This node exposes that cloud as a set of n8n operations: list recordings, fetch transcripts and summaries, upload new audio, manage tags and devices, and react to new content via a polling trigger.

n8n is a fair-code licensed workflow automation platform.

Installation
Operations
Credentials
Compatibility
Usage
Resources

⚠️ Unofficial / undocumented API. Plaud does not publish an API. This node was built by reverse-engineering the network traffic of web.plaud.ai. Endpoints, payloads, and the auth flow may change without notice. Use at your own risk and review Plaud's Terms of Service before deploying.

Installation

Follow the installation guide in the n8n community nodes documentation. The package name is n8n-nodes-plaud-cloud.

For self-hosted iterative development against a remote n8n container, this repo includes scripts/deploy.sh:

PLAUD_VPS=root@your-vps-ip ./scripts/deploy.sh

Operations

Plaud (action node)

| Resource | Operations | |---|---| | Recording | Get Many, Get, Update Filename, Trash, Delete, Get Download URL, Download Audio (binary) | | Transcript | Get (returns the transcript JSON for a recording) | | Summary | Get (returns the AI-generated summary as markdown) | | Account | Get Profile | | Upload | Upload Audio (multipart presigned-S3 flow) | | Tag | Get Many | | Device | Get Many (list devices linked to the account) |

Plaud Trigger (polling)

| Event | What it fires on | |---|---| | New Recording | A recording with a version_ms higher than the cursor (anything created or updated since last poll) | | New Transcript Available | A recording that has just transitioned from is_trans: false to is_trans: true |

The first poll silently establishes a cursor — installing the trigger does not flood you with historical recordings. State (cursor + seen-ids) is persisted in workflowStaticData('node').

Credentials

Plaud has no API-key endpoint, and the web login encrypts the password client-side with a public key. Reproducing that from a generic HTTP client requires reverse-engineering obfuscated JS, so this node uses a different mechanism: a long-lived bearer JWT extracted from a logged-in browser session.

Prerequisites:

• A Plaud account with at least one recording in the library.
• Access to a browser with DevTools (any modern desktop browser).

Steps to obtain the credential value:

1. Log in to web.plaud.ai.
2. Open DevTools → Network tab.
3. Click any request to api-*.plaud.ai (e.g. the request to /user/me).
4. Find the authorization request header.
5. Copy its value without the leading bearer prefix.
6. In n8n, create a new Plaud API credential and paste the value into Access Token (JWT).

The regional API host is detected automatically from the token's region claim, so there is no second field to configure. The token typically lasts approximately ten months.

Compatibility

• Tested against n8n 1.74.0 and later.
• Requires Node.js 18 or later in the n8n runtime.
• Known regions auto-mapped: aws:eu-central-1, aws:eu-west-1, aws:us-east-1, aws:us-east-2, aws:us-west-1, aws:us-west-2, aws:ap-southeast-1, aws:ap-southeast-2, aws:ap-northeast-1, aws:ap-south-1. Other regions fall back to api.plaud.ai (Plaud's discovery host); please open an issue with your region claim if you hit this fallback so the mapping can be extended.

Usage

Auto-transcribe-and-post

Plaud Trigger (Event: New Transcript Available)
  ↓
Plaud (Resource: Transcript, Operation: Get, Recording ID: ={{ $json.id }})
  ↓
Slack / Email / Notion ... (consume $json.transcript)

Archive recordings to S3

Plaud Trigger (Event: New Recording)
  ↓
Plaud (Resource: Recording, Operation: Download Audio, Recording ID: ={{ $json.id }})
  ↓
AWS S3 (Upload Binary Data)

Upload an external recording into Plaud

HTTP Request (fetch some .ogg)
  ↓
Plaud (Resource: Upload, Operation: Upload Audio, Filename: "Meeting 2026-01-15")

Known limitations

• Audio download URLs expire after 1 hour. Fetch immediately if you need the binary.
• Summary content is gzipped Markdown server-side; the node decompresses transparently.
• Upload uses the multipart chunk split returned by Plaud's /file/get_upload_presigned_url — Plaud decides how many parts. Voice-memo-sized files work fine; very large uploads (>1 GB) are untested.
• Plaud has no published rate limit. Polling at one-minute intervals is safe; sub-minute polling is untested.

Resources

• n8n community nodes documentation
• Plaud website
• Reverse-engineered API notes (this repository)

License

MIT