Groq switch cut API cost 30%. Production stays on gpt-4o. Monthly cap set at $30.
Offline transcription.
Free. Unlimited.
On-device speech-to-text, straight to Markdown
No network needed. A local Whisper model handles it all.
Transcription runs locally. 、
、 Yet meeting tools still want a monthly subscription.
Whisper runs on any machine. Yet every meeting-notes tool ships your audio to the cloud, charges $20/month, and locks output in a proprietary format. If you're a developer, that should feel wrong.
Sound familiar?
- Local-capable task, monthly subscription Runs on your CPU, but they charge for cloud
- Output is PDF or proprietary Won't open in Obsidian
- Audio uploaded to their cloud Your conversations, their servers
- No offline mode No network, no transcription
Your PC does the work
On-device transcription. No network required.
- ✓Install and go. No API key, no account needed
- ✓Output is Markdown. Obsidian, Git, whatever
- ✓Audio stays local. Never leaves your machine
- ✓Switch to cloud only when you need top-tier accuracy
With Reki,meeting notes work differently
Install it. Done.
An on-device Whisper model transcribes right on your machine. No account, no API key. Completely free with no usage limits.
Saved as .md. Do what you want with it.
Drop it into your Obsidian vault. Push it to Git. Grep it. Whatever you do with Markdown, it just works. No proprietary format locking you in.
1. Default to on-device transcription
2. Use cloud (Scribe) only for critical meetings
3. Monthly cost cap set at $30
Your audio never leaves your machine.
This is what makes Reki fundamentally different. Audio stays on your PC. Transcription runs on a local model. Nothing crosses the network. We never see the contents of your meetings.
- TodayEngineering StandupTranscription config optimized...
- 3d agoTech ReviewGraphQL migration PoC plan
- Last weekArchitecture ReviewMicroservice split decision
- 2w ago1:1 / Tanaka & MatsushimaHiring plan revised
- Last monthInfra PlanningSupabase migration decided
Any meeting tool. As-is.
OS-level audio capture. No bot invite needed.
SaaS vs DIY vsReki.
| Features | Reki note | SaaS (Otter / Fireflies etc.) | DIY Script / CLI |
|---|---|---|---|
| Offline transcription | On-device, unlimited | Cloud only, usage limits | Build Whisper yourself |
| Output format | Markdown | Proprietary | You implement it |
| Data storage | Local (no relay) | Cloud required | You configure it |
* Conceptual comparison
Offline by default.
Cloud when you need it.
- ✓ On-device transcription is free and unlimited
- ✓ Cloud mode uses top-tier models like Scribe
- ✓ Zero CPU load — processing happens off your machine
Install. Hit record.Get .md files.
Tanaka: Auth protocol decided — going with OAuth 2.0. GraphQL migration PoC starts next week.
Matsushima: DB schema changes are involved, so I'll run the migration script tests first.
Suzuki: Blue/Green deploy environment will be ready by Friday.
Meeting name, attendees, and tool auto-linked
Search across all past meetings
"What did we decide last week?" — answered
.md per meeting. Searchable decision log
Tanaka: Auth protocol decided — going with OAuth 2.0. GraphQL migration PoC starts next week.
Matsushima: DB schema changes are involved, so I'll run the migration script tests first.
Suzuki: Blue/Green deploy environment will be ready by Friday.
Meeting name, attendees, and tool auto-linked
Search across all past meetings
"What did we decide last week?" — answered
.md per meeting. Searchable decision log
Free unlimited transcription.
Go Pro only when you want to.
Data stays on your machine. Output is Markdown.
Auth protocol: OAuth 2.0 decided. GraphQL migration PoC starts next week. DB migration script test first.
Q3 hiring plan revised. One engineer hire instead of two. Remaining budget goes to AI infrastructure.
Groq switch cut API cost 30%. Production stays on gpt-4o. Monthly cap set at $30.
Auth protocol: OAuth 2.0 decided. GraphQL migration PoC starts next week. DB migration script test first.
Q3 hiring plan revised. One engineer hire instead of two. Remaining budget goes to AI infrastructure.