Notes with Ollama
Echo writes notes (a summary, key decisions and action items) from a finished transcript, using Ollama on your Mac. Every note links back to the passages it came from, so you can check it.
Set up Ollama
- Install Ollama from ollama.com/download and start it.
- In Echo, open Models → Local notes and choose Check connection.
- Download a recommended model from the same page.
- Select an installed model as the default there, or in Settings → General → Default local notes model.
Then open a meeting with a transcript, go to the Notes tab and generate notes. You can also pick a different installed model in the meeting’s Details.
Choosing a model
Echo recommends three small models that run well on an Apple Silicon laptop:
| Model | Download | Notes |
|---|---|---|
qwen2.5:3b (default) | about 1.9 GB | A good balance of speed and quality |
llama3.2:3b | about 2.0 GB | Similar size; try it if you prefer its style |
gemma3:4b | about 3.3 GB | Slower, sometimes more thorough |
Any model you have pulled in Ollama shows up in the list. Larger models give better notes but need more memory and time; as a rough guide, a model needs free memory of about its download size plus a couple of gigabytes.
Connection
Echo talks to Ollama at http://127.0.0.1:11434 by default. Change it under Settings → General → Local notes connection. Only loopback addresses are accepted, so notes never go to another machine.
Review the output
Language models make things up. Notes are marked as an AI draft; check them against the linked passages before relying on them. You can edit notes, and every edit creates a new version.