01
Describe the bug
Say what broke, when it started, and what must stay unchanged. DictaFlow removes the verbal cleanup.
Voice prompting for Claude, ChatGPT, Cursor, VS Code, Codex, terminals, and GitHub
Hold a key and explain the full task. DictaFlow cleans up the speech and types it where your cursor is, with the filenames, APIs, libraries, and constraints intact.
2,000 words free each month. No credit card required. Test it on one real task.
You said:
Um, can you fix the auth thing? Actually, the refresh token race in useSession.ts. Keep the existing API contract, add a regression test, and don't rebuild the client...
DictaFlow types:
Fix the refresh-token race in useSession.ts. Preserve the existing API contract, add a regression test, and avoid rebuilding the client.
Where voice earns its keep
AI coding tools are only as useful as the prompt. The weak point is often the handoff from your head to the prompt box.
01
Say what broke, when it started, and what must stay unchanged. DictaFlow removes the verbal cleanup.
02
Include file names, APIs, test expectations, and release limits without turning the prompt into a typing project.
03
The prompt lands as editable text before you send it. Fix a name or add one missing detail, then hand it to the agent.
04
Use the same input layer in Claude, ChatGPT, Cursor, the terminal, GitHub issues, Slack, and docs.
Good fit
Know the boundary
Use DictaFlow when you want to put clear text into an app. If you need voice commands for navigation, refactoring, or hands-free computer control, pair it with Talon, Cursorless, Voice Control, or another command layer.
That split works better than asking one tool to do both jobs badly.
Five-minute setup
DictaFlow already handles common programming language and software terms. A short setup makes project-specific words much more reliable.
Set your work domain so cleanup expects APIs, architecture language, debugging detail, and coding terminology.
Add repository names, internal services, package names, and casing-sensitive terms such as GitHub, TypeScript, PostgreSQL, and useSession.
Put the cursor in Claude, ChatGPT, Cursor, VS Code, a terminal, or GitHub. Hold the DictaFlow hotkey, speak, then release.
Use the same bug or feature you planned to work on today. Judge the full prompt-to-result loop, not a canned sentence.
One layer across the stack
Claude has voice features. ChatGPT has voice features. Some editors have speech extensions. Those can be useful, but they stop at the app boundary.
DictaFlow follows the cursor. Start the task in Claude, clarify it in GitHub, answer a review in Slack, and write the release note in your editor with the same hotkey.
A practical free test
You get 2,000 words free each month. No card. Install it, choose Software Engineering, and talk through one real task.
Pro is $7/month or $69/year if it earns a place in your workflow.
FAQ
Yes. DictaFlow types into the active text field in Claude, ChatGPT, browsers, and desktop apps on supported platforms.
You can dictate short snippets, comments, variable names, and code edits. DictaFlow is strongest for natural-language prompts, documentation, and review text. A keyboard is still better for dense symbol-heavy code.
DictaFlow handles common software terms and lets you protect exact project vocabulary. Add names, acronyms, libraries, and preferred casing once, then reuse them across apps.
No. DictaFlow creates editable text in the prompt box. You review it and choose when to send it.
Use one real task. Compare how much context you give, how many corrections you make, and whether the agent needs fewer follow-up prompts.