Miko Assistant

Some developers had started building software just by talking to their terminals. How does that work? Miko Assistant is our in-house macOS app for finding out. It sits between engineers and Claude Code and listens for commands.

The Question

AI coding assistants have accelerated development enormously, but working with them still mostly means typing. We wanted to know whether a developer can collaborate with AI entirely through voice and stay productive while trying to keep in mind, fewer context switches, faster ideation and a more natural conversation with the tools.

What We Built

To support this experiment, we developed Miko Assistant, an internal macOS application designed specifically for AI-assisted software development. Miko Assistant acts as the bridge between the developer and AI coding agents and manages multiple development sessions while enabling hands-free interaction.

Voice Coding Integration

One of the key capabilities of Miko Assistant is its voice coding workflow. The application captures spoken instructions through Speech-to-Text (STT), converts them into structured prompts, and forwards them directly to Claude Code. Developers can describe features, request refactoring, explain bugs, or ask implementation questions entirely through voice. Miko Assistant can also read Claude's responses aloud in it's own voice, so the exchange stays conversational.

Multi-Session Project Management

Software engineers rarely work on a single task throughout the day. Miko Assistant supports spawning multiple independent Claude Code sessions, allowing developers to work across different repositories or projects simultaneously. Each session maintains its own context, making it easy to switch between tasks without losing conversation history or development state. This creates a more scalable workflow for engineers who regularly juggle multiple projects.

Looking Ahead

Miko started as an experiment and has stuck around as our internal platform for testing how AI reshapes engineering work. Voice is the first step; autonomous development workflows, smarter project orchestration, and deeper toolchain integration are next.