Work

Private AI workflow

A Private Client

I built a private AI workflow that turns a client's recordings into organized notes and action items, with sensitive audio handled by AI on the client's own machine.

Private capture workspace with a locally transcribed conversation. Decisions and action items sit beside the recording list.
0 bytessensitive audio sent to cloud servicesArchitectural boundary: sensitive audio is transcribed locally.
Client
Private business operations
Role
Built
Timeline
2026 · Delivered
Stack
n8n, Whisper large-v3 (local, CUDA), FFmpeg, GPT-4o, Google Drive API, Asana, Railway

AI turns recordings and screenshots into organized notes and tasks in the client’s existing tools, with daily and weekly summaries.

Voice memos, calls, meetings, and screenshots arrived everywhere, with no reliable filing path. Sensitive audio could not go to cloud transcription.

Useful information was captured inconsistently and decisions had to be remembered manually.

  1. Built 13 automated workflows that collect recordings and screenshots, then use AI to transcribe the audio, tag each capture, pull out action items and decisions, and file the results.
  2. Kept sensitive audio on the client’s own computer, where a local AI model transcribes and classifies it.
  3. Added AI-written daily and weekly summaries.

Make privacy the architecture

Run the sensitive transcription and classification lane locally, rather than sending audio away to determine whether it is sensitive.

Use the systems already in place

Route each capture into the owner’s existing project structure instead of creating a second filing habit.

What the project delivered

2,351recordings and screenshots filedDrive metadata, 9–24 June 2026.
520tasks createdRecorded in Asana, 8 June–15 July 2026. This counts tasks created, not completed.

Say hello.

Tell me what you'd like to build or improve.

Quick navigation

Find a page.

Use the arrow keys to choose a result, Enter to open it, or Escape to close.