About
Hey! π I'm Nikolas, a software engineer, and I built Yapper because I was tired of typing walls of text into AI tools.
You know the drill: detailed prompts for Claude, context for Cursor, PR descriptions, code reviews, Slack threads explaining architecture decisions... My hands were dying.
Now I just talk. And it's saving me hours every day.
π€ The problem I solved for myself
As devs, we're typing more than ever:
π Long prompts for coding agents (Claude, Cursor, Copilot)
π¬ Explaining context to ChatGPT/Claude
π PR descriptions and code review comments
π Documentation and technical specs
πΌ Slack walls of text to explain "why we did it this way"
I realized I was spending more time typing to AI than actually coding. My RSI was getting worse. Something had to change.
β‘ How Yapper works
1οΈβ£ Press Option + Space (or your custom shortcut)
2οΈβ£ Speak naturally
3οΈβ£ Release - your words appear wherever your cursor is
That's it. Works in VS Code, Cursor, terminal, browser, Slack - anywhere you can type.
π Built for developers who care about privacy
- 100% Offline - Your voice never leaves your Mac. Runs on Apple's Neural Engine using WhisperKit. No cloud, no API calls, no data collection.
- No subscription - $24 once, yours forever. No monthly fees.
- BYO API keys - Optional AI enhancement uses YOUR Gemini/OpenAI/Anthropic/xAI keys. You control costs, you own the data.
I built this because I didn't want to send my voice (potentially containing proprietary code discussions) to random servers.
π οΈ Features that matter
- Multiple Whisper models - Tiny (fastest) to Large v3 Turbo (most accurate). Pick your speed/accuracy tradeoff.
- Custom vocabulary - Add technical terms, library names, your codebase's weird naming conventions. `kubectl`, `Kubernetes`, `NextJS`, your company's internal toolsβYapper learns them.
- Toggle or hold-to-record - Quick tap to toggle, or hold while speaking. Whatever fits your flow.
- File transcription - Transcribe meeting recordings, standups, or voice memos. MP3, WAV, M4A.
- Transcript history - Search past transcriptions. Great for finding "what did I say in that prompt yesterday?"
π‘ Real use cases from my daily workflow
Talking to coding agents:
> "Refactor this function to use async/await instead of callbacks, maintain the same error handling behavior, and add TypeScript types for the return value"
PR descriptions:
> "This PR adds rate limiting to the API endpoints. I've implemented a token bucket algorithm with Redis as the backend. The main changes are in the middleware folder. Breaking change: the rate limit headers are now lowercase."
Code review comments:
> "This looks good but I'm concerned about the N+1 query here. Consider using a join or batch loading the related records."
Slack technical discussions:
> "The reason we went with PostgreSQL over MongoDB for this service is that we need strong consistency guarantees for the payment flow, and the relational model fits our data better."
All of these: spoken in seconds, not typed in minutes.
π Launch Special: 50% Off
- Regular price: $49
- Launch price: $24 (one-time)
- β Includes all future updates
- β Priority support (I actually respond)
π§ Tech details for the curious
- Built in Swift/SwiftUI
- Runs as a menu bar app (no dock clutter)
- ~75MB to ~1.5GB for models
- Requires macOS 14+, Apple Silicon recommended
π¬ Happy to answer questions!
What's your biggest pain point with typing? I'm actively developing Yapper and prioritizing features based on feedback from early adopters.
Stop typing prompts. Start speaking them. ποΈ