Skip to main content
Vaani
Engineering & QAPreset: Developer Preset

Spoken Bug Reports → Issues

Turn spoken observations into structured, reproducible GitHub and Linear issues.

The Workflow Challenge

Engineers and QA leads spend 15 to 20 minutes manually formatting repro steps, expected behaviors, and stack traces after spotting bugs during local test runs. Dictating observations while clicking through the app captures precise context without interrupting your investigation flow.

Designed for and currently being tested in GitHub Issues, Linear, and Jira web interfaces on macOS.
Speech-to-Text TransformationIllustrative simulation
Spoken Audio Input (Unedited)

“hey um so basically when users with expired refresh tokens hit the billing page it gives them a 500 internal server error instead of redirecting them to login so we should fix that in next sprint”

Disfluencies detected · Conversational phrasing
Refined Output (Developer Preset)

“Users with expired refresh tokens encounter HTTP 500 on /billing instead of receiving an automated redirect to /login. Schedule remediation for the next sprint.”

Ready for active window insertion

How it works in 3 steps

01

Speak while clicking through repro steps

Hold Option+Space while testing your build and narrate observed errors, API codes, and expected behaviors without switching windows.

02

Developer preset extracts technical nouns

Vaani preserves status codes (HTTP 500), file paths, endpoints (/billing), and strips conversational filler words.

03

Inserts clean Markdown directly into your issue tracker

Release the key and the formatted problem statement is pasted straight into the description field with native clipboard fallback.

Common questions

Does Vaani understand code identifiers and API status codes?

Yes. The Developer refinement preset is tuned to preserve technical syntax, camelCase identifiers, HTTP codes, and stack trace references.

Can I add custom team acronyms and service names?

Yes. The custom dictionary allows you to teach Vaani proprietary microservice names, internal acronyms, and repo names.

Is my proprietary code or bug detail sent to public cloud servers?

When transcription is fully local, audio stays on your device. In cloud mode, audio is sent to the provider you configure.

Early Access

Try the Spoken Bug Reports → Issues Workflow in Early Access

Join the waitlist to receive access to macOS prototype builds equipped with this refinement preset.

No marketing trackers or advertising cookies. Unsubscribe anytime.