Voice-Controlled Coding: Using Wispr Flow with Claude Code and Beam
What if you could build software by just talking? With the right tool stack, you can. This guide shows you how to combine Wispr Flow (voice dictation), Claude Code (AI coding assistant), and Beam (terminal organizer) for a seamless, hands-free development workflow.
The tool stack
Wispr Flow
Voice-to-text for macOS that works anywhere you can type. Speak naturally and it transcribes in real-time. Perfect for dictating prompts to Claude Code.
Claude Code
Anthropic's CLI for Claude. It can read your codebase, write code, run commands, and iterate on changes – all from your terminal.
Beam
Terminal organizer that lets you run multiple Claude Code sessions in separate workspaces. One for frontend, one for backend, one for tests – all organized and searchable.
Why this combination works
Each tool solves a different problem:
- Wispr Flow eliminates typing – you just speak your intent
- Claude Code translates your intent into actual code changes
- Beam keeps multiple AI sessions organized so you don't lose track
Together, you get a workflow where you can say "add a logout button to the navbar that clears the session and redirects to the login page" and watch it happen – while keeping your hands free for coffee, notes, or just thinking.
Setting it up
1. Install the tools
- Wispr Flow: Download from wisprflow.com
- Claude Code:
npm install -g @anthropic-ai/claude-code - Beam: Download from getbeam.dev
2. Organize your workspaces in Beam
Create a workspace for each part of your project:
- ⌘N to create a new workspace
- Double-click the title bar to rename it (e.g., "Frontend Claude", "Backend Claude")
- Run
claudein each workspace to start a session - ⌘S to save your layout for next time
3. Configure Wispr Flow
Wispr Flow works anywhere you can type, including the terminal. A few tips:
- Use the default hotkey (usually Option twice) to start/stop dictation
- Speak naturally – Wispr handles punctuation and formatting
- For code-specific terms, you may need to spell out unusual variable names
The workflow in action
Open Beam and restore your layout
Hit ⌘P to open the quick switcher, type the name of your saved layout, and restore your workspace instantly.
Navigate to the right Claude session
Use ⌘P again to jump to "Frontend Claude" or "Backend Claude" – wherever you need to make changes.
Activate Wispr Flow and speak your intent
Press your Wispr hotkey and say something like: "Add form validation to the email field. It should check for a valid email format and show an error message below the input if invalid."
Let Claude Code do its thing
Claude reads your codebase, finds the right files, and makes the changes. Review what it did, approve or ask for adjustments – all by voice if you want.
Switch contexts seamlessly
Need to work on the backend now? ⌘P → "Backend Claude" → speak your next request. Each session maintains its own context.
Tips for voice-controlled coding
Be specific with your requests
Instead of "fix the bug", say "fix the null pointer exception in the user authentication function when the email field is empty". Claude Code works better with context.
Use Beam's workspaces for different concerns
Keep your frontend Claude session separate from backend. When you switch contexts, you don't have to re-explain the project structure – each session remembers.
Save your layout
Once you have your Claude Code sessions organized in Beam, save the layout (⌘S). Tomorrow you can restore it instantly and pick up where you left off.
Who is this for?
This workflow is ideal for:
- Developers with RSI or carpal tunnel – reduce typing strain while staying productive
- Multitaskers – dictate code changes while reviewing designs or documentation
- Fast thinkers – speaking is often faster than typing, especially for complex requests
- Anyone who wants to try a new way of coding – it's surprisingly fun once you get used to it
Limitations to know about
Voice coding isn't perfect for everything:
- Unusual variable names may require spelling out or correction
- Complex code review is still easier with eyes on the screen and hands on keyboard
- Noisy environments can affect dictation accuracy
That said, for high-level direction ("add this feature", "refactor this function", "write tests for this module"), voice + Claude Code is remarkably effective.
Try it yourself
The beauty of this stack is that each tool is useful on its own. You don't have to go all-in on voice coding immediately. Start by organizing your Claude Code sessions in Beam. Then try dictating a few prompts with Wispr Flow. See if it clicks for you.
Many developers are surprised how natural it feels after a few days. And once you've experienced asking for a feature and watching it get built while your hands are free? It's hard to go back.