Start with prompt templates designed for interruptible voice, screen-aware explanations, camera context, and short spoken answers. Use them now as session briefs and later as API-ready inputs.
Prompt template
Draft a session brief
Copy-ready GPT Live prompts
These prompts are session briefs, not proof of live API access. They are written for short spoken turns, interruptions, and future voice/screen context.
Explain a product or workflow while the user shares a screen.
Run a short voice support session and produce a clean next step.
Hold a natural conversation while correcting important mistakes.
Explain camera or image context without overclaiming.
Pick the closest task, then adjust the context, voice behavior, and ending format.
Watch my screen, explain only the current step, pause when I interrupt, and summarize the next action before moving on.
Listen to the issue, ask one clarifying question at a time, give short spoken fixes, and end with a support-note summary.
Hold a natural conversation, correct major mistakes gently, repeat improved phrasing, and keep the conversation moving.
Teach through questions, check understanding often, use brief examples, and recap the lesson in plain language.
Use camera or shared image context, describe what is visible, ask before guessing, and explain one observation at a time.
Answer the real-time question briefly, name uncertainty when needed, suggest one follow-up search, and finish with a decision-ready summary.
Voice prompts work best when the model knows the job, the input modes, and the response rhythm.
Say what GPT Live is helping with and what kind of assistant it should be during the session.
Name the inputs it can use: speech, screen, camera, shared media, search context, or user corrections.
Keep answers short, allow interruptions, ask when context is missing, and specify the final summary.
Open the homepage entry, paste a template, and keep the brief ready for future GPT-Live API wiring.