Yakki.ai – Say it. Ship it. Increase your output with your voice

Name: Yakki.ai – Say it. Ship it. Increase your output with your voice
Availability: InStock
Author: imazio

by imazio·Mar 2, 2026·3 points·0 comments

Visit Project View on HN

AI Analysis

●●SolidSolve My ProblemDark HorseSlick

Local dictation with speaker diarization beats Whisper.cpp, but crowded against built-in macOS.

Strengths

•Speaker diarization (who said what) is genuinely non-obvious and useful for meetings, rare in local transcription.
•True on-device processing with sub-200ms latency and automatic context awareness (adjusts output to app) shows real polish.
•Audio capture from any app (Teams, Zoom, podcasts) without cloud is a meaningful privacy advantage over competing tools.

Weaknesses

•macOS-only severely limits addressable market when Windows, iOS, and web alternatives exist or are easier to build.
•"Local dictation app" is crowded (Whisper.cpp, built-in macOS Voice Control, Speeko, Otter.ai offline); no clear differentiation beyond speaker diarization.

Post Description

I built a macOS menu bar app for capturing and leveraging the audio you produce/generate.

Hold fn, talk, release, your text appears wherever your cursor is. Transcription runs on-device using local models, works offline, and nothing leaves your machine if you do not want to.

I wanted a faster dictation app that was put me in control, but also let me do things like recording my meetings without having to invite any weird guess to the Teams call, generate perfect meeting notes with insights, and allow me to go back to who said what when.

It can also help transcribe my podcasts when I don’t have time to listen to it all, then generate insights and key points.

The focus is on what to do with what you capture, not just perfect dictation.

I have yet to start marketing it and I am struggling with positioning it in a saturated market. Any advice or feedback will be SUPER welcome!