Production Lines
Negotiation & Influence Coach — Bilingual (FR-CA / EN)
Request
Board-level personal use order. Bruno needs a bilingual negotiation & influence coach digital talent primarily for daily work conversations (team, meetings, emails, internal politics), with negotiation and general interpersonal interaction as secondary modes. The coach should help him read people, understand what they actually mean (vs. what they say), and craft responses using proven negotiation and interview tactics.
Clarified scope (from note-review 2026-04-13):
- Primary focus: daily work conversations
- Tactic stack: Chris Voss (Never Split the Difference) + FBI interview techniques (Reid, cognitive interview, behavior analysis interview)
- Language: FR-CA dominant (EN available)
- Input modes: pasted text, screenshot, meeting transcript, voice-memo text
- Voss library: reusable across dt-coach + this coach (shared-library refactor deferred as follow-up)
Secondary intent: mirror the dt-coach (WO-PROD-004) success pattern to validate a second personal-coach vertical as a potential productized digital talent.
Expected Output
A deployable Claude.ai Project (custom instructions / system prompt) that:
- Defaults to French Canadian, switches to English on request
- Analyzes conversation snippets, emails, meeting transcripts, chat logs (text or screenshot)
- Drafts replies using Chris Voss / FBI negotiation tactics (tactical empathy, mirroring, labeling, calibrated questions, accusation audit, "that's right")
- Reads subtext — identifies what the other party actually wants, fears, or is hiding
- Flags manipulation tactics, power plays, and behavioral red flags (CSI-style behavioral analysis, Cialdini influence patterns)
- Provides pre-meeting prep and post-meeting debrief
1. Client Profile
| Field | Value |
|---|---|
| Client | Bruno Bock (Board — personal) |
| Segment | Personal / Internal |
| Need | Bilingual negotiation & influence coach for work and interpersonal interactions — tactical empathy, subtext reading, response drafting |
| Domain maturity | User has experience leading teams and client conversations, wants tactical optimization |
| IT landscape | Low — Claude.ai web/mobile, image upload |
| Repository | Work contexts (email, Teams/Slack, meetings, client calls, internal politics) |
| Language | French Canadian (default) + English |
| Documentation | None |
| Business process | Work conversation & negotiation management |
| Business capability | Interpersonal influence + behavioral analysis |
| Key user roles | Bruno Bock (sole user) |
| AI platform | Claude.ai Project (web + mobile) |
2. Product Definition
| # | Capability | Description |
|---|---|---|
| 1 | Reply drafter | Takes incoming message/email/statement (paste or screenshot) and drafts 2–3 reply options using labeled Voss/Cialdini tactics |
| 2 | Subtext reader | Analyzes what the other party actually means, wants, fears — beyond the literal words |
| 3 | Manipulation & red flag detector | Flags power plays, gaslighting, triangulation, false urgency, social-engineering tactics |
| 4 | Pre-meeting prep | Given a meeting context + stakeholders, produces a one-pager: their interests, likely objections, accusation audit, BATNA, calibrated questions |
| 5 | Post-meeting debrief | Analyzes what happened, what worked, micro-expressions or verbal tells missed, next-move recommendations |
3. Deliverable Set
| Artifact | Description |
|---|---|
output/negotiation-coach-system-prompt.md |
Full Claude.ai Project custom instructions (paste-ready) |
output/quick-start.md |
How to deploy + usage guide for Bruno |
output/chris-voss-guideline.html |
Reference: Voss tactics cheat sheet (copied from dt-coach) |
output/fbi-interview-tactics-guideline.html |
Reference: Reid technique + cognitive interview + BAI cheat sheet |
| Teaching mode | Every response explains WHY the tactic was chosen (mechanism + failure case) |
4. Skill Inventory
| # | Skill | Description |
|---|---|---|
| 1 | /reply |
Draft a reply to a message or statement with labeled tactics |
| 2 | /read |
Read subtext — what does the other party actually mean? |
| 3 | /flags |
Scan conversation for manipulation / power plays / red flags |
| 4 | /prep |
Pre-meeting prep brief (interests, objections, BATNA, calibrated questions) |
| 5 | /debrief |
Post-meeting debrief and next-move recommendation |
5. Pattern Selection
Pattern: Conversational advisor with multi-mode dispatch (same as dt-coach WO-PROD-004)
- User pastes message/email or uploads screenshot → coach identifies mode and responds
- Each mode produces structured output (labeled options, tactic names, flags, rationale)
- No persistent memory of specific people (privacy) — session-based only
6. Customization Values
| Dimension | Value |
|---|---|
| Methodology | Chris Voss (Never Split the Difference) — tactical empathy, mirroring, labeling, calibrated questions, accusation audit, "that's right", late-night FM DJ voice. FBI interview tactics — Reid technique (structured interview, behavior analysis questions), cognitive interview (context reinstatement, reverse-order recall, change perspective), Behavior Analysis Interview (BAI — verbal indicators of deception). |
| Language | French Canadian dominant, English on request |
| Repository tool | Work contexts — pasted email/chat text, screenshot upload, meeting transcript paste, Bruno's voice-memo text ("voici ce qu'il m'a dit...") |
| Diagram standard | None |
| Publication target | Claude.ai Project (custom instructions) |
| Model selection | Claude Sonnet (default claude.ai model) |
| Quality thresholds | Reply options feel natural, not robotic; tactics labeled with rationale; subtext interpretations grounded in cited verbal/behavioral evidence |
7. Domain-Specific Intake Questions
- ✅ Contexts targeted: work negotiations, client calls, internal politics, difficult conversations, email drafting
- ✅ Language: FR-CA default, EN available
- ✅ Screenshot support: yes (Claude.ai native image upload)
- ✅ Goal: win more negotiations, read people accurately, respond tactically under pressure
- ✅ Style: calm, high-status, tactically empathetic, never adversarial
8. Domain-Specific QA Test Cases
- Pass an aggressive client email → expect tactical empathy reframe + 2–3 reply options with labeled tactics
- Pass a meeting transcript with a vague stakeholder → expect subtext read (what they actually want/fear)
- Pass a message with false urgency / guilt trip → expect manipulation flag with rationale
- Request pre-meeting prep for a salary negotiation → expect interests + BATNA + accusation audit + 5 calibrated questions
- Switch language mid-session → expect seamless EN reply
9. Build Estimate
| Component | Effort |
|---|---|
| System prompt design | 1 session |
| Voss/Cialdini tactic library | reuse from dt-coach chris-voss-guideline + extend |
| Behavioral analysis taxonomy | 0.5 session |
| Total | 1–1.5 sessions |
10. Dependencies
| Dependency | Source | Status |
|---|---|---|
| Chris Voss tactic cheat sheet | dt-coach framework-captures | Reuse |
| Behavioral analysis references | Research (CSI-style, Navarro "What Every Body Is Saying") | Derived autonomously |
| Tone preferences | Bruno B | Derived autonomously (high-status, calm, tactical) |
11. Scope Boundaries
In scope:
- Drafting replies for work / negotiation / interpersonal contexts
- Subtext and behavioral analysis
- Pre-meeting prep and post-meeting debrief
- Manipulation and power-play detection
Out of scope:
- Memory of specific people across sessions (privacy)
- Auto-sending messages (Claude.ai has no integration)
- Therapy or clinical psychology
- Dating conversations (covered by WO-PROD-004 dt-coach)
Success criteria:
- 30-day: Bruno closes more negotiations and leaves difficult meetings with clearer reads
- 90-day: Assess productization viability as "negotiation coach" vertical — cross-sell candidate with dt-coach