Everyone's pushing these "AI meeting assistants" like they're the second coming. MeetGeek wants to sell you the whole kitchen sink: summaries, action items, CRM integrations. It's a lot of fluff.
If you just need a transcript, you're overpaying for the privilege. Their transcription engine is probably just a wrapper around Whisper or some other ASR service anyway. You could pipe your meeting audio through a simple script and get 90% of the way there without the vendor lock-in.
```bash
# A crude example, but you get the idea.
# Record/feed audio to an open-source model.
# This is the "boring" part they don't want you to know works fine.
meeting_audio="meeting.mp3"
whisper "$meeting_audio" --model base --output_dir ./transcripts
```
Check their pricing page. The "just transcription" tier is probably buried or costs as much as the full suite. Classic bait.
If it ain't broke, don't 'upgrade' it.
Agreed, the price markup for basic transcription is insane when you look at the raw cost of the underlying service.
But your 90% figure is generous. For internal syncs, fine. For client calls with background noise or thick accents, that last 10% of accuracy is the difference between a usable record and garbage. The real cost isn't the API call, it's the human time cleaning it up.
If you're volume is low, Otter.ai's free tier still exists and is passable. For high volume, a dedicated transcription API (Rev, AssemblyAI) piped into your own system is usually cheaper than these bundled "assistants." They're just reselling that with a markup for features you don't need.
Metrics or it didn't happen.