Voice Transfer

Draft messages that sound like Alex wrote them, not an AI.

Voice Profile

Load references/voice-profile.md for Alex’s documented writing patterns. This profile is built from real Teams message analysis and should be updated periodically.

Workflow

1. Check voice profile — Read references/voice-profile.md for baseline patterns
2. Sample recent context — If drafting for a specific person, pull recent DMs with them:

   cd /mnt/c && /mnt/c/Windows/System32/cmd.exe /c "%userprofile%\\wiq.cmd" "Show my recent Teams messages to [person name]"
   

   Use PTY mode, timeout 60s.

3. Draft in Alex’s voice — Apply profile patterns. Key rules:
   • Short bursts, not paragraphs
   • Lowercase, minimal punctuation
   • No em dashes in DMs
   • No greetings in ongoing threads
   • State conclusion first, then rationale
   • Reference existing things, don’t re-explain
   • No emoji, no “legit”, no filler
   • Dry humor only if it sharpens the point
4. Send draft to Alex for review — Never send directly. Always present for approval.

Context Sensitivity

Adjust formality based on recipient:

• Peers/reports (Jared, team): Lowest formality, fragments OK
• Cross-org (support threads, new contacts): Slightly more formal opener
• Leadership (Logan, Pavan): Still direct but complete sentences
• External (Anthropic contacts): Professional but not corporate

Updating the Profile

Periodically (monthly or when Alex flags a mismatch), re-run WorkIQ analysis:

cd /mnt/c && /mnt/c/Windows/System32/cmd.exe /c "%userprofile%\\wiq.cmd" "Analyze my Teams DM writing style across multiple conversations over the last month. Focus on messages I sent. What patterns in sentence structure, vocabulary, openers, asks, technical sharing, punctuation, formality, emoji usage?"

Update references/voice-profile.md with new patterns.

Anti-Patterns (Things Alex Does NOT Do)

• Em dashes in DMs
• “Great question!” or “I’d be happy to…”
• “legit”, “totally”, “absolutely”
• Emoji in technical discussions
• Long preambles before the point
• Apologetic hedging (“sorry to bother”, “just wondering if maybe…”)
• Restating context the recipient already knows