2026-08-28·by Sijie Wang#market#awareness#communication

rapport-and-presence

The mostly-{live} cluster of communication: real-time conversation, nerves, body, eye contact. It matters here because it is precisely the human residual of cold-start and engagement-is-a-graded-dial's primary-interaction cells — the person doing the genuine relationship work that an agent can't. Captured so that residual is skilled, not improvised. A few pieces (written conversational structure, agent-voiceover (VO) tone/register) transfer.

Rapport & conversation {live}{written}

  • Match, Mirror, Lead (MML) — mirror the other's vocal foundations (rate, volume, energy, melody) AND body first — fast similarity = instant safety; don't blast a quiet/grieving person with your loud default ("you emotionally punch them in the face"). Once in rapport, gradually lead them up. You can only lead a state after joining it.
  • High-Low-Buffalo — each person shares a High (going great), Low (deliberate vulnerability), Buffalo (something interesting). Two people = 6 threads. Manufactures topics + seeds the vulnerability that fast-tracks depth; kills the weather-loop. Frame it with permission + go first.
  • Conversational threading — catch the specific emotionally-live words the other drops; follow with "who specifically / why / what's the most valuable lesson"; when answering, hand them threads to pull. "Most conversations die because there are no threads." Small talk is the on-ramp to big talk — don't skip it, don't loop in it.
  • 3 follow-ups before common ground — when they share a peak win, ask three follow-ups about their peak before "I have one too" — converts one-upping into connection.
  • The 4 wants (diagnose then deliver) {live}{written} — detect their top priority and answer in that register: Information (flat → give data) · Connection (emotion → answer with emotion + story) · Peace (conflict-averse → no drama) · Fun (playful → play back).
  • Color read (Erikson's 4 colors) {agent-VO for the register}{live for the read} — Red = results/direct · Yellow = fun/social · Green = peace/calm · Blue = logic/precise; deliver in their language. "Be formless like water."
  • The proof-of-listening stack — reflect back a word or the emotion they named (the only proof you listened) · pause 2 extra beats (they add something deeper) · follow up. Listen in four channels: ears (words), heart (mood), eyes (body), vocal changes ("yeah it's okay" = not okay).
  • "Yes, and" not "Yes, but" · negative-feedback sequence (private, permission, framed as growth, warm, end positive) · say the name 3× + a specific goodbye · Coaching-Habit loop for serial complainers ("what's on your mind → tell me more → what else → what's the main problem → how can I help").

Nerves & confidence {live} (mindset {written})

  • Audience-conscious, not self-conscious — nerves = self-focus; redirect all attention outward ("I'm here to serve them"). Attention is finite; flood it with the audience and there's no capacity left for self-monitoring. For agent content this is the transferable core: write to serve the viewer, not to look impressive.
  • Burn off the adrenaline — discharge before going on (push-ups / brisk walk): "the only way your voice shakes is if your body shakes."
  • Box / Wim Hof breathing · in-the-moment reset (pause → breath → slow the rate) · ⭐ relabel fear as excitement (identical body signals, only the label differs — say "this is excitement").
  • Look-Feel-Sound cheat — of the three, feel is hardest yet everyone fixes it first; control the two easy ones (look: upright, take up space, 2mm smile; sound: volume, melody, the pause) and the brain back-fills the feeling. "Unfamiliar, not unnatural" — a bigger gesture that feels wrong needs reps, it isn't wrong for you (he recanted "fake it till you make it" → "explore the unfamiliar").
  • Graded exposure — climb the trigger ladder (seated toast → stand → team meeting); rejection therapy (collect 20 NOs). "We mistake unfamiliarity for fear."
  • The 2mm smile — carry a slight smile (not a grin); most default to a resting frown. "That 2mm change may be the most profound change you make."

Body, gesture, eye contact {live}

  • Gestures = "subtitles for your voice" — enumerate in space, contrast in opposite hand-spaces (his viral "10/10 technical vs 3/10 communication" bit worked because people saw it), punctuate to land a point. Gesture when talking, still when listening.
  • Eye-contact don'ts — don't dart, don't scan-and-glance, don't go mechanical left-center-right, don't look over their heads (the "1cm above the lens" advice is F-tier), don't lock on one person. Real, distributed, present contact.
  • Step forward on the peak (loud→soft pull-in), move only with purpose, bigger gestures → more presence + more voice, stand for calls / hands above the table / door-frame posture reset.
  • Virtual specifics — look into the camera, hide self-view, appear in social space (head-to-waist), cameras ON (off → slump → dead voice), amp energy +50–80% (the lens drains it).

Parent: communication · the machine-usable slice: what-transfers-to-agent-content.

about this entry

One of sijie's wiki entries. The AI on this site is grounded in the same corpus and answers in sijie's voice, with citations back to entries like this one — answering costs sijie money, so it waits behind a code: enter an access code →