Anthropic has introduced a voice mode feature for Claude Code, allowing developers to interact with the AI coding assistant via spoken commands rather than typing.
The feature is currently available to about 5% of users, according to Anthropic engineer Thariq Shihipar, with a broader rollout expected in the coming weeks.
Voice commands for coding tasks
With the new voice mode, developers can start the feature by typing “/voice” and then speaking instructions to the system.
For example, a user could say “refactor the authentication middleware,” and Claude Code will process the request. Developers can also ask the AI to review code, debug issues, generate documentation, or perform other programming tasks.
The goal is to create a more natural, hands-free workflow that allows engineers to interact with the AI assistant while working across multiple screens or reviewing code.
More conversational development workflow
Instead of switching between typing and reading responses, programmers can speak instructions and receive answers in a conversational format.
This approach could support AI-assisted pair programming, where developers collaborate with AI systems in real time to write or improve code.
However, Anthropic has not yet shared some technical details about the feature. It remains unclear whether the tool has usage limits, latency constraints, or environment-specific restrictions.
The company has also not confirmed whether the voice technology was developed internally or relies on external providers.
Expansion of Claude’s voice capabilities
Anthropic had already introduced voice mode for its Claude chatbot in 2025, allowing users to hold spoken conversations for everyday tasks.
Extending voice capabilities into Claude Code signals a broader strategy to make AI tools more interactive. Instead of acting only as text-based assistants, AI systems are gradually becoming conversational collaborators for developers.
Voice features may be especially useful when engineers are multitasking, reviewing architecture diagrams, or debugging complex systems.
How does voice mode work?
Voice mode allows users to speak to Claude and hear spoken responses instead of typing and reading text. The feature can run in two main modes:
- Hands-free mode, where Claude listens continuously and responds after natural pauses in speech
- Push-to-talk mode, where users hold a button while speaking to reduce background noise
Users can switch between text and voice during the same conversation, and the system keeps the full context of the discussion.
Voice conversations are saved as text transcripts in chat history, and they count toward normal usage limits based on a user’s subscription plan.
Anthropic says the voice system includes safeguards to prevent misuse. The feature only offers a limited set of preset voices, which prevents voice cloning or impersonation.
The company also said the AI generates responses rather than copying existing speech patterns, and standard safety monitoring remains active during voice interactions.
