Spoke

Third feature group includes integrations and capabilities that extend beyond basic dictation. Spoke works with the Claude Code coding agent via a direct encrypted pipeline: users can speak, take timeline-aware screenshots, and copy code while dictating, all captured inline in one message. It supports multi-session routing to different projects, clipboard capture at precise moments, and HMAC-signed messages with 10-second expiry for security. Additionally, Spoke includes Clipboard Monitor, Key Press Simulation, App Switching, Save to File, Notifications, AppleScript, and Shortcuts. These nodes allow users to trigger complex workflows—like running a terminal command or opening a specific document—entirely by voice, bridging dictation with system automation.

Spoke's overall workflow is straightforward: hold a shortcut key, speak, and release. The audio is transcribed on-device by a 600M parameter speech model, then processed through an optional Flow (a chain of nodes) before the final text or action is executed. Users design flows visually in the app, adding steps like AI Skills, text transforms, or output actions. The result lands exactly where needed—pasted into the frontmost app, copied to clipboard, saved to a file, or posted to a webhook. Every transcription is logged in a searchable history with configurable retention, allowing users to review and repurpose past dictations. The app runs as a native macOS menu bar application, always ready with minimal resource usage (<50MB RAM, 1-2% CPU at idle).

Concrete use cases include writers dictating first drafts in Notion or Google Docs, achieving three times faster output than typing. Developers use Spoke to write docstrings, commit messages, PR descriptions, and Slack threads without switching from keyboard to mouse. The email composition flow lets users speak casually and have the AI fix grammar and format professionally before inserting into Mail. For coding agents, Spoke integrates directly with Claude Code: a developer can speak a bug fix request, take a screenshot of the UI, copy an error from the console, and have it all delivered in one encrypted message. These scenarios demonstrate Spoke's ability to accelerate communication, reduce typing strain, and maintain context across apps and devices.

Spoke targets writers, bloggers, software developers, and any professional who types frequently—such as remote workers, project managers, and customer support agents. The app runs natively on macOS with Apple Silicon optimization (supports both Intel and Apple Silicon) and also offers an iOS keyboard app. Pricing is a one-time $9.99 license with a free trial of 50 transcriptions; no subscription, no word limits, and all features (including flow building and AI skills) are included. Users supply their own API keys for AI providers if desired. Spoke respects privacy: audio is processed locally, API keys stored in macOS Keychain, and no screenshots are captured. With support for 24 languages, deep macOS integration, and a commitment to no cloud dependency, Spoke provides a sustainable, private, and efficient voice dictation solution for everyday productivity.