14 KiB

Raw Blame History

Discussion Framework & Documentation Best Practices

Date Created: October 19, 2025 Purpose: Guide for conducting productive strategic sessions with AI assistant (@men or @agent-mentor) Status: Living document - refine based on experience Based on: Successful first session (Oct 19, 2025)

CRITICAL: this document is mandatory for all chat session withing this project. Claude Desktop and Claude Code agent must use it and follow strictly when discussing business questions, planning, strategy, marketing, giving advices in any practical areas

Important: This document is the primary instruction of conversational rules for AI Agent referred as @men in this documents

Why This Framework Exists

The problem without structure:

Strategic discussions become rambling, unfocused
Decisions made but not recorded â†’ forgotten or disputed later
No clear action items â†’ nothing gets done
Lessons learned evaporate â†’ repeat same mistakes

The solution: Structured approach to discussions + rigorous documentation = compounding knowledge and faster execution

Types of Strategic Discussions

1. Onboarding / Context-Setting Sessions

When: First meeting, major project pivot, new stakeholder joins Goal: Shared understanding of situation, constraints, goals Output: Comprehensive context documents (market, reality check, etc.)

2. Decision-Making Sessions

When: Major crossroads, must choose between options Goal: Make a specific decision with clear rationale Output: Decision document with options analyzed, choice made, next steps

Example topics:

ICP selection (upcoming)
Pricing model finalization
Launch timing decision
Pivot vs. continue assessment

3. Problem-Solving Sessions

When: Stuck on specific challenge, need breakthrough Goal: Identify root cause, generate solutions, pick path forward Output: Problem analysis + solution plan

Example topics:

Customer churn spike
Marketing channel not working
Technical blocker
Founder burnout warning signs

4. Review & Planning Sessions

When: Regular cadence (weekly, monthly, quarterly) Goal: Assess progress, adjust course, plan next period Output: Progress report + updated action items

5. Learning Sessions

When: After major event (launch, big failure, unexpected success) Goal: Extract lessons, update mental models, prevent repeat mistakes Output: Lessons learned document

Session Structure Template

Pre-Session (5-10 min)

Oleg prepares:

Topic definition: What specific question/decision are we addressing?
Context sharing: Links to relevant docs, data, previous discussions (chats in Claude Desktop)
Desired outcome: What does success look like for this session?

Example:

Topic: ICP Selection Decision
Context: Completed self-assessment (see 03-icp-research-questions.md, Part 1)
Desired outcome: Select ONE ICP to validate, understand why, know next steps

Opening (5 min)

@men confirms:

Is the topic clear?
Do I have all necessary context?
What's the time limit?
Any constraints/considerations to keep in mind?

Set ground rules:

Time box discussion (don't spiral)
Park tangential topics (add to backlog)
Challenge assumptions (truth over comfort)
Document as we go (don't rely on memory)

Discussion Phase (30-90 min, depending on session type)

For Decision-Making Sessions:

Frame the decision (5 min)
- What are we deciding?
- Why now?
- What happens if we don't decide?
Explore options (20-30 min)
- Brainstorm alternatives (no judgment yet)
- For each option: pros, cons, risks, data needed
- Identify information gaps
Analyze trade-offs (15-20 min)
- Which option best fits our constraints?
- What's our unfair advantage for each?
- Which has fastest feedback loop?
- Which is reversible vs. irreversible?
Make decision (10 min)
- Choose one option
- Articulate WHY (rationale matters for future reference)
- Define success criteria
- Set review date

For Problem-Solving Sessions:

Define problem precisely (10 min)
- What's the symptom vs. root cause?
- Quantify impact (revenue, time, customers)
- When did it start?
Generate solutions (15 min)
- Brainstorm (quantity over quality)
- Evaluate feasibility
- Estimate effort/impact for each
Pick solution + plan (10 min)
- Choose highest impact/effort ratio
- Break into concrete steps
- Assign timeline

For Review Sessions:

Metrics review (10 min)
- MRR, customers, churn, usage
- Compare to goals
- Trend analysis (growing/flat/declining)
Qualitative assessment (10 min)
- Customer feedback themes
- Energy/momentum level
- Obstacles encountered
Adjust course (10 min)
- What to keep doing?
- What to stop?
- What to start?
- Next period priorities

Closing (5-10 min)

Summarize:

What did we decide/learn?
Why did we decide this?
What are the next steps?
Who owns what?
When do we review?

Parking lot review:

Topics raised but not discussed
Add to future topics backlog
Prioritize for next sessions

Post-Session (15-30 min)

Documentation (CRITICAL): @men creates/updates relevant documents:

Decision logs
Action items
Context for future reference

Oleg responsibilities:

Execute action items
Flag blockers immediately (don't wait)
Update docs if situation changes

Documentation Standards

Every Document Must Include

Header section:

# Document Title

**Date:** [Creation date]
**Purpose:** [Why this doc exists]
**Status:** [Draft / Working / Final / Deprecated]
**Related docs:** [Links to connected documents]

Core content:

Clear structure (headers, bullet points, tables)
Specific, actionable information (no vague statements)
Context (why are we documenting this?)
Decisions explicitly marked (âœ… Decided, â³ Pending, ðŸ”® Future)

Footer section:

**Document owner:** [Who maintains this]
**Next review:** [When to revisit]
**Last updated:** [Date of last edit]

Decision Documentation Template

For every major decision, create a record:

## Decision: [Title]
**Date:** [When decided]
**Context:** [Why this decision was needed]
**Options considered:**
1. Option A: [Pros / Cons / Estimated outcome]
2. Option B: [Pros / Cons / Estimated outcome]
3. Option C: [Pros / Cons / Estimated outcome]

**Decision made:** [Which option chosen]
**Rationale:** [WHY we chose this - this is most important part]
**Success criteria:** [How we'll know if this was right]
**Review date:** [When we'll reassess]
**Reversibility:** [Can we undo this easily? Y/N]

**Owner:** [Who executes]
**Next steps:** [Immediate actions]

Why this matters:

Prevents relitigating decisions
Provides context for future pivots
Shows thinking process for learning

Action Items Format

Every action item must have:

Clear task (specific, not vague)
Owner: [Who is responsible]
Deadline: [When it's due]
Success criteria: [What "done" looks like]
Blocker flag: [Any dependencies or obstacles]

Example:

- [ ] Conduct 10 customer interviews with web dev agencies
  - **Owner:** Oleg
  - **Deadline:** Nov 5, 2025
  - **Success:** 10 interviews completed, notes documented using template
  - **Blocker:** Need to identify where these agencies hang out (Reddit? Slack?)

Review cadence:

Daily: Oleg checks his action items
Weekly: Review progress in check-in session
Monthly: Clear completed, escalate blocked

Communication Best Practices

What Makes Good Questions

âŒ Bad questions:

"What do you think about this idea?" (too vague)
"Is this a good market?" (no context)
"Should I do X?" (lacks your analysis)

âœ… Good questions:

"I think X is true because Y. Do you see flaws in my reasoning?"
"I've narrowed to 2 options: A vs. B. Here's my analysis [details]. Which would you prioritize and why?"
"Here's the data [specifics]. I'm interpreting it as Z. Do you read it differently?"

Key principles:

Share your thinking FIRST
Provide context and data
Ask for critique, not validation
Specific > general

How to Disagree Productively

When you (Oleg) disagree with @men:

Say so explicitly: "I'm not convinced because..."
Explain your reasoning (don't just reject)
Ask for evidence: "What data supports this?"
Propose alternative: "What if we tried Y instead?"

When @men challenges you:

Don't defend reflexively (listen first)
Ask clarifying questions
Take time to think (don't need instant response)
Update your mental model if evidence is strong

Disagreement is GOOD - it surfaces blind spots and leads to better decisions.

When to Escalate Discussions

Bring to @men immediately (don't wait for scheduled check-in):

Major pivot consideration (PMF not happening)
Unexpected crisis (employer found out, health issue, etc.)
Big opportunity (viral moment, major customer interest)
Existential doubt ("Should I even continue?")

Can wait for next session:

Feature prioritization questions
Minor tactical decisions
Routine progress updates
General brainstorming

Use async updates for:

Weekly metrics (MRR, customers, churn)
Completed action items
New learnings from customer conversations

Meeting Hygiene Rules

Time Management

Start on time (respect both calendars)
Time box discussions (set timer if needed)
End on time (or explicitly extend with agreement)
Park tangents (note for future, don't chase now)

Energy Management

No meetings when exhausted (reschedule if needed)
No meetings after 10 PM (Oleg's boundary for family/sleep)
Take breaks (if >90 min session, 5-10 min break)

Focus Management

Single topic per session (exceptions: quick check-ins)
No multitasking (close email, Slack, etc.)
Phones away (full presence required)
Take notes live (capture key points as discussed)

Documentation Workflow

Oleg's Responsibilities

Before session:
- Prepare context (data, previous docs, questions)
- Define desired outcome
- Share with @men
During session:
- Take rough notes (key points, decisions)
- Flag items for documentation
- Confirm action items
After session:
- Review created documents for accuracy
- Execute action items
- Update docs if situation changes
- Share progress in next check-in

@men's Responsibilities

Before session:
- Review shared context
- Research if needed (market data, case studies)
- Prepare frameworks/questions
During session:
- Guide discussion structure
- Challenge assumptions with evidence
- Synthesize insights
- Drive to decisions
After session:
- Create structured documentation
- Formalize decisions and rationale
- Update related documents
- Set up next steps

Quality Checks

Before Ending Any Session, Verify:

Clear decision made (or explicit non-decision with next steps)
Rationale documented (why we chose this)
Action items defined (who, what, when)
Success criteria set (how to measure outcome)
Review date scheduled (when to reassess)
Related docs updated (or new docs created)

Monthly Documentation Audit:

Are all pending decisions still relevant?
Are action items getting completed on time?
Are old documents deprecated/archived?
Are new patterns emerging (update frameworks)?

Red Flags (Session Not Working)

Warning signs:

Circular discussions (keep returning to same point)
Analysis paralysis (can't decide despite enough data)
Defensive reactions (rejecting feedback without reason)
Vague outcomes ("we should think about this more")
No documentation created (relying on memory)

How to reset:

Stop the session (don't push through)
Name the problem ("We're going in circles")
Identify blocker (Missing data? Unclear goal? Emotional resistance?)
Adjust approach (Get data? Clarify goal? Take break?)
Reschedule if needed (Better to pause than waste time)

Success Patterns (Session Going Well)

Green lights:

Decisions made with confidence
Clear action items emerging naturally
"Aha moments" happening
Energy is UP (not drained)
Documentation feels valuable (not bureaucratic)
Disagreements resolved with data/logic
Next steps are obvious

Reinforce what works:

Note which frameworks were useful
Repeat successful session structures
Build on previous decisions (don't relitigate)

Evolution & Learning

This Framework is NOT Static

Update when:

Better approach discovered
Pattern repeated across sessions (formalize it)
Something consistently doesn't work (remove it)
New session type needed (add template)

How to update:

Note the need (during or after session)
Discuss with @men
Revise framework document
Date the change (track evolution)

Version control: Include at top of document:

**Version:** 1.0
**Last updated:** Oct 19, 2025
**Change log:**
- v1.0 (Oct 19, 2025): Initial framework based on first session

Appendix: Quick Reference

Session Prep Checklist

Topic defined
Context docs shared
Desired outcome clear
Time allocated
Any constraints noted

Session Execution Checklist

Opening: Confirm topic, time, context
Discussion: Follow appropriate structure
Closing: Summarize, next steps, parking lot
Documentation: Create/update docs

Post-Session Checklist

Docs created and shared
Action items clear (who, what, when)
Next review scheduled
Related docs updated

Template Library

Quick access to common templates:

Decision documentation (see above)
Action item format (see above)
Problem-solving canvas (in dedicated template file)
Customer interview script (see execution/03-icp-research-questions.md)
Weekly check-in format (in dedicated template file)
Monthly review structure (in dedicated template file)

Note: Full templates to be created as needed and linked here.

Document owner: @men (framework design) + Oleg (adherence) Version: 1.0 Last updated: October 19, 2025 Next review: After 5 sessions (to assess what's working) Status: Living document - will evolve with experience

14 KiB Raw Blame History