Automating meeting transcripts/summaries

I’m trying to fully automate the process of recording an in-person meeting, transcribing it, summarising it with an LLM, and receiving a formatted summary via email. Most of the pipeline is working fine - once I have the transcript, Claude processes it, reformats it, and emails me the final result. The issue is getting the transcript automatically in a way that triggers the automation.

Initially I used Otter.ai, which works well for transcriptions, but automation is a nightmare.

  • The Pro plan doesn’t allow any automation.
  • The Business plan can monitor a Dropbox folder for new recordings and auto-transcribe them, but it doesn’t put the transcript back in that folder, so I can’t trigger the next step.
  • Otter does have an API, but it’s locked behind the Enterprise plan, which requires contacting sales. Not viable for a small business with a couple of employees.

I looked at Rev.com, which offers an API on their automated transcription plan, but I’m running into issues:

  • The API works for checking new orders, but when I try to retrieve the transcript, it throws an error instead of returning the text.
  • First-line support couldn’t help, and they’ve escalated it to their API team, but no resolution yet.

At this point, I’m considering:

  1. Finding another transcription service that actually works with API-driven automation without enterprise pricing.
  2. Scraping the transcript from Otter as a last resort, though I’d rather avoid this.
  3. Building my own transcription pipeline using Whisper.cpp or something similar. I tested Whisper a while ago, and it was okay but not great – has it improved? Would it be reliable enough for automated meeting notes?

This should be a solved problem – automatically transcribing meetings and emailing a summary isn’t rocket science. But every existing solution either lacks automation or gates API access behind enterprise plans.

Does anyone know of a transcription service with solid automation options that actually works? Or has anyone built their own setup for this? Open to suggestions.

(Edit: clarify that it’s in-person meetings)