The original is one click away. Open original ↗
Using ChatGPT voice mode to rehearse high-stakes conversations
Executive overview
Most people use ChatGPT's advanced voice mode for trivial questions, but it can serve as a low-stakes rehearsal space for sales calls, executive presentations, and personal reflection. By assigning the AI a specific persona and inserting a trigger phrase to pause for coaching, users get real-time feedback without risking a live opportunity. The therapy use case extends this further: a structured brain-dump followed by iterative Socratic questions yields surprisingly targeted personal insight. Connecting a years-long digital journal to a large-context model like Gemini pushes the depth further still.
Practicing difficult conversations with AI before they happen converts preparation time into compounding performance gains.
Sales call rehearsal
- Use a separate AI model (ChatGPT, Claude, DeepSeek) to generate a set of buyer personas before starting the voice session.
- Five useful archetypes: analytical buyer, skeptical challenger, relationship-driven buyer, price-conscious negotiator, time-pressed executive.
- Prompt the voice AI to adopt one persona and begin a live sales conversation.
- Insert a nonsense trigger phrase ("purple unicorn") to break character and request immediate coaching mid-conversation.
- After feedback, say the trigger again to resume the scenario exactly where it left off.
- Constrain the scenario with real context: named buyer type, time limit, and pre-meeting background to make the simulation more accurate.
- Repeat across multiple personas to stress-test messaging against different objections.
Presentation and negotiation prep
- Executive presentations and board briefings routinely attract skeptics, visionaries, and analysts who will push back hard.
- The old approach — writing out rebuttals alone — is slow and misses the verbal dimension of defending a hypothesis.
- With voice mode, assign specific debate personas and practice verbalising counterarguments out loud before the real meeting.
- The same trigger-phrase technique lets you pause, get critique, and refine your argument in a single session.
- Low-stakes repetition in AI rehearsal means fewer surprises and less hesitation when the stakes are real.
AI-assisted therapy and self-reflection
- Start with an uninterrupted brain-dump: instruct the AI not to speak until you say the trigger phrase, preserving natural thinking pauses.
- Assign a persona grounded in a specific therapeutic philosophy or public thinker (the video uses Jordan Peterson as an example) to anchor the style of questioning.
- Request iterative questions rather than immediate answers: each response from you should reshape the next question, building a progressively deeper picture of the issue.
- This Socratic loop uncovers root causes more effectively than a one-shot advice dump.
- Advanced option: export years of digital journals (e.g., from Obsidian) into a large text file and upload to a model with a large context window; the AI then references your actual history when probing.
- This approach surfaces patterns in thinking and behaviour that are invisible in any single conversation.
Practical tips and constraints
- The mobile ChatGPT app is more reliable for voice mode than the browser version.
- Walking with the AI in headphones extends session length naturally and reduces self-consciousness.
- Trigger words should be unusual phrases unlikely to appear in normal conversation ("purple rhino," "purple unicorn").
- Constraints sharpen simulations: specify the buyer, time box, and any background context the AI should treat as established fact.
- Sessions can run 30–45 minutes; treat them like deliberate practice, not casual chatting.
More like this — when you're ready for early access.
Join the waitlist for a personal account and content recommendations based on what you're working on.
No spam. Unsubscribe at any time.
You're on the list. We'll be in touch before launch.