Configuration
API keys, model choices, language hint, and microphone selection.
Everything is configured from the Settings view inside the app. Nothing is hardcoded, and no key ever needs to be committed anywhere.
API keys
Ogma talks to three services, each with its own key:
| Key | Used for | Where to get it |
|---|---|---|
| OpenAI | Whisper transcription | platform.openai.com/api-keys |
| Anthropic | Speaker labels + meeting notes (Claude) | console.anthropic.com |
| Notion (optional) | Syncing meetings to your workspace | notion.so/my-integrations — create an internal integration |
Leave the Notion fields empty to skip Notion sync entirely; everything else still works.
How keys are stored
On save, keys are written to the OS keychain (Windows Credential Manager, macOS Keychain) and blanked in the on-disk config.json. If no native keystore is available, or a keychain write fails, Ogma falls back to keeping the key in config.json rather than losing it — so a key you saved is never silently dropped.
Models
Two model settings are exposed:
- Whisper model — defaults to
whisper-1 - Notes model — the Claude model used for speaker attribution and notes, defaults to
claude-sonnet-5
You normally don't need to touch these; they exist so you can adopt newer models without waiting for an app update.
Language hint
Whisper auto-detects the spoken language by default. If your meetings are consistently in one language, set the language field to an ISO code (e.g. en, es) — it improves accuracy and avoids misdetection on noisy openings.
Microphone
Ogma records from the system default input device unless you pick a specific one under Input device. Choose the physical mic closest to the room conversation — a laptop mic in the middle of a table generally beats a headset for multi-speaker capture.
Notion connection
See Notion sync for connecting an integration and creating or linking a Meetings database.