tertiusDictation saves time only when the full loop is dependable: start, speak, stop, process, review, and place. A fast transcript that lands in the wrong field or loses the original thought can create more work than typing.
1. Start without changing applications
A global shortcut lets the recording begin while the cursor remains in email, a document, a project tool, or a browser. Tertius exposes quick capture from Windows and lets the user configure capture behavior rather than requiring a separate editor for every thought.
2. Decide what “clean” means before speaking
Different messages need different finishes. A quick reply may only need filler removal and punctuation. A client update may need a subject, short paragraphs, and a next step. Meeting reflection may need decisions, risks, and actions.
Tertius separates the transcript from the outcome. You choose a mode or reusable playbook before capture, then the fixed pipeline transcribes the audio and applies the selected transformation. This is not a promise that the result is always ready to send; important text still deserves review.
Clean transcript
Readable speech with filler and false starts reduced while keeping the original meaning.
Send-ready email
A subject suggestion, greeting, concise body, and explicit next step.
Action items
Concrete tasks with owners and dates only when they were actually stated.
Custom playbook
Your own instructions, structure, and tone paired with a repeatable outcome.
3. Control where the result goes
Automatic paste is convenient when the active field remains stable. Clipboard copy is safer when an application changes focus or the text needs review first. Tertius keeps copying and auto-paste as independent settings, enabled by default, so testers can turn off either behavior without losing the other.
For sensitive or irreversible fields, disable auto-paste and review the copied result before insertion. Never rely on generated text for legal, medical, financial, or safety-critical decisions without qualified review.
4. Teach the tool your terminology
Names, acronyms, product terms, and industry vocabulary are common failure points. A personal lexicon gives the transcription workflow better context. Test the exact terminology you use in real rooms and with your real microphone; generic demo sentences do not reveal these issues.
5. Make recovery explicit
Good recovery includes visible recording state, a cancellable operation, a retained result, and clear error feedback. Tertius can save transcript text, the finished outcome, and local metadata in optional Trail history. It does not store the original audio or video in Trail.
Privacy and connectivity
Tertius is not offline dictation. The desktop sends authenticated requests to the managed Tertius service, which keeps the OpenAI credential on the server and applies monthly quotas, request limits, and concurrency controls. Testers sign in with their approved email and a six-digit code; they do not supply an OpenAI key. Review the privacy summary before using confidential material.
Common questions
Does Tertius work on Windows 10 and Windows 11?
The private beta targets 64-bit Windows 10 and Windows 11.
Can it paste dictation into the active field?
Yes. Copying and automatic paste are independent settings and are enabled by default.
Does it work offline?
No. Transcription and outcome processing use the managed Tertius online service.
What does the free tester account include?
It currently includes 120 transcription minutes per month plus a managed outcome-processing allowance. Limits may change as the private beta is measured.