One workflow I’ve been using AI for recently isn’t exactly an automation, but it has removed my dependency on several paid tools.
For a long time, I used AI meeting note-takers to record and transcribe Google Meet calls. They worked well, but most of them require a monthly subscription. More importantly, I didn’t like the idea of all my meeting transcripts living on someone else’s servers. I wanted everything to stay on my own system.
So I asked Claude Code to help me build a local transcription workflow.
The setup was surprisingly straightforward. I installed Python, the Faster Whisper model (which suited my laptop’s configuration), and FFmpeg so it could process video recordings from Google Meet.
With AI’s help, I wrote a simple Python script that runs from the terminal. Whenever I have a meeting recording, I pass the video file to the script, and it generates a transcript directly on my computer.
From there, I can use Claude Code again to analyze the transcript. Depending on what I need, it can generate meeting notes, summarize discussions, extract action items, or audit the conversation.
The best part is that everything happens locally. There are no recurring subscription costs, and I have complete ownership of my transcripts.
It’s a small workflow, but it has made my note-taking process cheaper, more private, and easier to customize.
You can find out a few more basic uses of AI in my work here.