Automate the end-to-end flow from recording on iPhone to Obsidian-ready Markdown in Google Drive, using on-device transcription and optional local summarization.
The AI agent captures meeting audio from iOS Voice Memos, transcribes it on-device, and optionally summarizes the result. It then converts the transcript into clean Markdown via an n8n webhook. Finally, it uploads the Markdown file to a Google Drive folder that is synced with your Obsidian vault, delivering notes that are easy to search and link.
End-to-end automation from recording to Obsidian-ready notes.
Record audio in Voice Memos.
Transcribe locally on-device.
Optionally summarize with ChatGPT or a replaceable local model.
Convert transcript to Markdown via the n8n webhook.
Upload the Markdown file to Google Drive.
Sync the Markdown file with Obsidian vault.
This AI agent replaces manual note-taking with a private, end-to-end automation from capture to Obsidian-ready notes. It handles transcription, optional summarization, Markdown formatting, and vault synchronization with minimal setup.
A simple 3-step system for non-technical users.
Record the meeting in the iOS Voice Memos app; the AI agent is triggered when the memo is saved.
Transcribe the memo locally using on-device transcription; if enabled, pass the text to a local model or ChatGPT for a concise summary.
Send transcript and optional summary to the n8n webhook which formats a Markdown note and uploads it to a Google Drive folder synced with Obsidian.
A realistic, end-to-end use case.
A sales manager records a 25-minute client meeting on an iPhone using Voice Memos. The on-device transcription completes in minutes, and a brief summary is produced by the local model. The system converts the notes to a Markdown file named with the date and client, then uploads it to Drive, syncing with the Obsidian vault so the notes appear in the correct project folder within minutes.
People who need private, reliable mobile-to-Obsidian note workflows.
Privacy-first client meeting notes captured on-device.
On-device processing preserves client confidentiality.
Fast capture of stakeholder discussions with consistent formatting.
Immediate, shareable meeting notes for follow-ups.
Structured notes aligned with project folders in Obsidian.
Offline transcripts for archive and citation in Obsidian.
Tools that run the AI agent’s end-to-end workflow.
Source audio for transcription.
Orchestrates transcription and optional summarize steps.
Provides an optional summarization of the transcript.
Formats the transcript into Markdown and triggers storage.
Uploads the Markdown file to a vault-synced folder.
Receives Markdown notes via Drive sync for immediate access.
Practical scenarios to expand coverage of this AI agent.
Common questions about the AI agent and its workflow.
Yes. Transcription happens on-device to preserve privacy by default. The optional summarization step can use either a local model or a cloud-based service like ChatGPT, depending on your configuration. Data never leaves your device for transcription, and only the Markdown payload is uploaded via the webhook when you choose to proceed with storage. You can swap in a local model for all steps if you want to avoid any cloud involvement altogether. Always ensure your iOS and shortcuts are configured according to your privacy preferences.
Yes. The summarization step is optional and can be routed through a local AI model instead of ChatGPT. This keeps the workflow fully offline if you prefer. The local model can be deployed in your environment and wired into the Shortcut's flow via the same webhook. Consider performance and memory constraints when selecting a local model. You can switch back to cloud summarization at any time without changing the main pipeline.
Longer transcripts may encounter reliability issues; the workflow notes a practical limit around ~1 hour per memo. If a memo extends beyond this, consider splitting recordings into shorter segments. You can also adjust the workflow to handle chunked transcripts and compile them into a single Markdown note. For critical scenarios, perform a quick verification and manual adjustment of filenames and headings post-upload.
Set up the n8n webhook to receive the transcript payload from the Shortcut and transform it into Markdown. Point the webhook to your n8n instance and configure a Google Drive action to upload to a specific folder mirrored to Obsidian. Ensure the Drive folder is synced with your Obsidian vault. Test with a short memo to confirm the Markdown structure and file naming conventions. You may tailor prompts and filenames to fit your project types.
Yes. The prompts used by the local or cloud summarization step can be customized via the Shortcut and prompt templates. You can adapt the tone, length, and structure of the summaries to match your note style. If you use a local model, you can modify the model's prompt handling in your environment. Regular adjustments help align outputs with your preferred Obsidian structure.
Notes are stored as Markdown files in a Google Drive folder that your Obsidian vault is synced with. The transcription itself stays on-device by default, and only the Markdown content is uploaded when you enable the save action. Access controls on Drive and your Obsidian vault determine who can view or edit the notes. If privacy is paramount, use a local-only summarization path and keep the entire workflow offline.
Yes. You can create separate Google Drive folders and corresponding Obsidian vaults for different projects or clients. The n8n workflow can be configured to route transcripts to the appropriate folder based on metadata like client name or project tag. This keeps notes organized and correctly linked to each vault. If you switch vaults, ensure the prompts and file naming conventions reflect the target context.
Automate the end-to-end flow from recording on iPhone to Obsidian-ready Markdown in Google Drive, using on-device transcription and optional local summarization.
