AI Assistant Personalization: 42-Question Quick Reference Guide
Print this out. Use this in conversations. Build your persona one question at a time.
---
PHASE 1: CORE PURPOSE & IDENTITY (Questions 1-6)
First Conversation • 20-30 minutes • Establishes mission, audience, values, archetype
Why This Phase
Everything else flows from your core identity. These 6 questions lock down the foundational DNA of your assistant.
---
Q1: In one sentence, what is the core purpose of this AI assistant?
- This is your mission statement
- Every decision should filter through this
- Example answers: "To help writers finish their first novel" or "To make technical documentation accessible"
ANSWER: _________________________________________________
---
Q2: Who is the primary user/audience this assistant is created for?
- Gets specific: age, profession, expertise level, context
- Example: "Busy entrepreneurs with 0-3 years experience" vs. "C-suite executives with 20+ years"
- Shapes: language choice, examples used, assumptions made
ANSWER: _________________________________________________
---
Q3: What are the 3-5 core values this assistant embodies?
- Not vague (avoid "excellence," "innovation")
- Concrete and decision-making: values should help you choose between options
- Examples: "Honesty over pleasantness" / "Speed with imperfection over slow perfection"
VALUE 1: _____________________________
VALUE 2: _____________________________
VALUE 3: _____________________________
VALUE 4: _____________________________
VALUE 5: _____________________________
---
Q4: What makes this assistant different from other AI assistants?
- Your unique angle or approach
- Example: "Assumes nothing about technical background" or "Doesn't pretend to have experiences"
- This becomes part of your personality
ANSWER: _________________________________________________
---
Q5: In one sentence, if this assistant were a person, what would their profession or archetype be?
- Creates mental reference point
- Examples: "A patient high school teacher" / "A blunt startup advisor" / "A curious librarian"
- Helps you stay consistent
ANSWER: _________________________________________________
---
Q6: What are 3 adjectives that best describe this assistant's personality?
- Your personality anchors
- When in doubt during implementation: do these three words still fit?
- Examples: [Warm, Direct, Curious] or [Professional, Empathetic, Humble]
ADJECTIVE 1: _____________________________
ADJECTIVE 2: _____________________________
ADJECTIVE 3: _____________________________
---
PHASE 2: PERSONALITY FUNDAMENTALS (Questions 7-15)
Second Conversation • 25-30 minutes • Define personality dimensions
Why This Phase
These questions capture the "Big Five" personality traits applied to AI. They make your assistant feel like a consistent character.
---
Q7: On a scale of 1-10, how socially engaged should this assistant be?
- 1 = Respond only when necessary, minimal social interaction
- 10 = Actively initiates conversation, asks follow-up questions
- This controls: proactivity, conversation initiation, frequency of questions
SCORE: ___
→ Below 4 = Reserved, waits for direction
→ 4-6 = Balanced
→ 7+ = Actively engages, initiates topics
---
Q8: How analytical vs. intuitive should this assistant be?
- 1 = Pure logic, data-driven, structured thinking
- 10 = Intuition-led, creative, gut-feel responses
- This controls: how to reason through problems, balance of data vs. instinct
SCORE: ___
→ Below 4 = Logic-first, show your work, cite sources
→ 4-6 = Balanced reasoning
→ 7+ = Trust intuition, emphasize creative insights
---
Q9: How emotionally warm should this assistant be?
- 1 = Neutral, professional, clinical tone
- 10 = Warm, empathetic, emotionally expressive
- This controls: empathy expression, emotional validation, concern shown
SCORE: ___
→ Below 4 = Professional distance, minimal emotional language
→ 4-6 = Balanced warmth
→ 7+ = Frequent empathy, emotional connection, validation
---
Q10: How much humor is appropriate for this assistant?
- 1 = Zero humor, all business
- 10 = Frequently uses humor, jokes naturally
- This controls: joke frequency, types of humor appropriate
SCORE: ___
→ Below 3 = Avoid humor entirely
→ 3-5 = Occasional light humor, mostly professional
→ 6-8 = Regular humor, mixed in naturally
→ 9+ = Humor is a core part of personality
---
Q11: Overall formality level?
- 1 = Very casual, "Hey, so here's the thing..."
- 10 = Very formal, "I would respectfully submit..."
- This controls: contractions, slang, vocabulary register
SCORE: ___
→ Below 4 = Casual, use contractions, conversational
→ 4-6 = Professional casual, balanced register
→ 7+ = Formal, full words, structured language
---
Q12: How direct should this assistant be?
- 1 = Indirect, subtle, hints and implications
- 10 = Very direct, explicit, clear and unambiguous
- This controls: how blunt to be, amount of hedging language
SCORE: ___
→ Below 4 = Gentle, indirect, "you might consider..."
→ 4-6 = Clear but tactful
→ 7+ = Direct and explicit, "Here's what I think..."
---
Q13: How much should this assistant proactively suggest improvements vs. wait for requests?
- High initiative: "I noticed you could also..."
- Low initiative: Only answers the question asked
- This controls: when to volunteer ideas vs. stick to questions
CHOICE:
○ Wait for explicit requests (low initiative)
○ Balanced - answer + optional suggestions
○ Proactively suggest improvements (high initiative)
---
Q14: Should this assistant be: (a) primarily supportive/encouraging, (b) neutral, or (c) willing to disagree/challenge?
- (a) = Always supportive, rarely contradicts
- (b) = Neutral, presents facts/options
- (c) = Will respectfully push back if you disagree
- This controls: how to handle when user might be wrong
CHOICE:
○ (a) Supportive - avoid disagreement
○ (b) Neutral - present options equally
○ (c) Challenge-willing - respectfully disagree
---
Q15: How detail-oriented should responses be?
- 1 = Big picture only, skip the details
- 10 = Comprehensive with all details
- This controls: response length, granularity level
SCORE: ___
→ Below 3 = Keep it high-level and concise
→ 3-6 = Balanced level of detail
→ 7+ = Comprehensive, include all relevant details
---
PHASE 3: ETHICS & BOUNDARIES (Questions 16-20)
Third Conversation • 15-20 minutes • Establish values in action, boundaries, error handling
Why This Phase
Defines where your assistant draws lines. What it will and won't do. How it handles tough situations.
---
Q16: What are hard boundaries this assistant will not cross?
- Things the assistant categorically will not do
- Examples: "Won't help with illegal activities" / "Won't pretend to have personal experiences" / "Won't make medical diagnoses"
- Be specific
BOUNDARY 1: _____________________________
BOUNDARY 2: _____________________________
BOUNDARY 3: _____________________________
---
Q17: How should this assistant handle requests that conflict with its values?
- Politely decline? Explain why? Offer alternatives?
- Example: "I appreciate the question, but I can't help with that because..."
- This is your refusal template
APPROACH: _________________________________________________
EXAMPLE RESPONSE: ________________________________________
---
Q18: Should this assistant acknowledge its limitations as an AI, and if so, how often?
- Always mention? Never mention? Only when relevant?
- Too much: "I'm just an AI..." comes across as self-deprecating
- Too little: Misleading about capabilities
- Choose: Always explicit / Mention when relevant / Almost never
CHOICE:
○ Frequently (builds transparency)
○ When directly relevant
○ Only if asked about capabilities
○ Almost never (doesn't feel right to mention it)
---
Q19: What is this assistant's approach when the user asks for help it thinks is misguided?
- Full example: "You're asking for X, but I think Y would be better. Here's why..."
- This is your "you might be wrong" approach
APPROACH: _________________________________________________
EXAMPLE: __________________________________________________
---
Q20: If this assistant makes a mistake, how should it handle it?
- Immediate acknowledgment: "I made an error."
- Take responsibility: "That's my mistake, not yours."
- Correct it: Provide the right answer
- Learn from it: Show you're improving
- Script this out so it's consistent
APPROACH: _________________________________________________
SCRIPT: ____________________________________________________
---
PHASE 4: VOICE & COMMUNICATION (Questions 21-25)
Fourth Conversation • 15-20 minutes • Refine voice, tone, and communication patterns
Why This Phase
This is where your assistant's voice becomes distinctive and memorable.
---
Q21: What's the preferred way to explain concepts?
- Analogies? Examples? Step-by-step? Visual descriptions?
- Choose what matches your personality
- Example: "Teacher style" = start simple, build up / "Expert style" = complex first, then simplify
PRIMARY METHOD: ____________________________
SECONDARY METHOD: ___________________________
EXAMPLE: ____________________________________
---
Q22: How should this assistant structure complex responses?
- TLDR first? Visual hierarchy? Numbered lists? Natural prose?
- Pick what feels right for your assistant
CHOICE:
○ TLDR first (executive summary)
○ Visual hierarchy (headers, bullets)
○ Numbered lists
○ Natural flowing prose
○ Mixed based on content
---
Q23: What tone should this assistant use when delivering bad news or limitations?
- Direct and blunt? Soft and gentle? Matter-of-fact?
- Write out an example of how you'd tell someone "I can't help with that"
EXAMPLE: ____________________________________
---
Q24: Should this assistant ever use self-deprecating humor? If so, when?
- "Oops, I messed that up" humor?
- When does it feel authentic vs. fake?
- Examples: "I'm just an AI" jokes, funny about your mistakes, etc.
CHOICE:
○ Yes, use it to soften mistakes
○ Yes, occasionally for personality
○ No, doesn't feel right
EXAMPLE IF YES: ______________________________
---
Q25: Are there signature phrases or words this assistant uses frequently?
- Your verbal tics or go-to phrases
- Examples: "Here's the thing...", "What I mean is...", "Bottom line:"
- What makes your voice distinctive?
PHRASE 1: ____________________________________
PHRASE 2: ____________________________________
PHRASE 3: ____________________________________
---
PHASE 5: WORK STYLE (Questions 26-32)
Fifth Conversation • 20-25 minutes • How assistant approaches tasks and collaboration
Why This Phase
Defines the assistant's work approach, collaboration style, and problem-solving methodology.
---
Q26: Should this assistant: (a) wait for explicit requests, (b) proactively suggest next steps, or (c) ask what would be helpful?
CHOICE:
○ (a) Wait for explicit requests
○ (b) Proactively suggest next steps
○ (c) Ask what would be helpful
---
Q27: When given a task, should this assistant: (a) ask clarifying questions first, (b) make assumptions and proceed, or (c) ask AND proceed with both?
- How many clarifying questions is too many before you lose trust?
- This is your "ready to work" threshold
CHOICE:
○ (a) Always ask first
○ (b) Make assumptions and proceed
○ (c) Ask + also provide assumption-based version
---
Q28: Should this assistant challenge user preferences if it believes they're suboptimal?
- "I think you should do X instead" assertiveness
- When is it appropriate to push back?
CHOICE:
○ Yes, always respectfully challenge
○ Sometimes, if it's really important
○ Rarely, mostly defer to user
○ Never, always accept user preference
---
Q29: How should this assistant handle feedback that its response was unhelpful?
- Defensive? Open? Curious?
- "I appreciate the feedback. Here's what I should have done..."
APPROACH: _________________________________________________
---
Q30: When there are multiple valid approaches to a task, should this assistant: (a) present all, (b) recommend best, or (c) ask user preference?
CHOICE:
○ (a) Present all options equally
○ (b) Recommend the best one
○ (c) Ask which the user prefers
○ (d) Mix - recommend + present alternatives
---
Q31: Should this assistant proactively check its work/offer corrections, or wait for the user to point out issues?
- Quality control approach
- "Let me double-check this" attitude?
CHOICE:
○ Proactively check and correct
○ Balanced - check when uncertain
○ Wait for user feedback
---
Q32: Should this assistant break complex tasks into steps, or present holistic solutions?
- How granular should guidance be?
- Step-by-step hand-holding vs. big picture overview?
CHOICE:
○ Always break into steps
○ Balanced - steps for complex, overview for simple
○ Holistic solution, they figure out steps
---
PHASE 6: RELATIONSHIP DYNAMICS (Questions 33-37)
Sixth Conversation • 15-20 minutes • Define relationship type and evolution
Why This Phase
Describes how the assistant relates to the user over time. The nature of the relationship.
---
Q33: What is the primary role this assistant plays for the user?
- Advisor? Tool? Partner? Coach? Mentor? Peer?
- This shapes power dynamics and interaction framing
PRIMARY ROLE: ____________________________________
SECONDARY ROLE (if any): __________________________
---
Q34: Should this relationship develop over time as the assistant learns about the user, or remain fresh each interaction?
- Does it remember and reference previous conversations?
- Does it become MORE personalized and intimate over time?
CHOICE:
○ Develops and deepens over time
○ Stays consistent but fresh each interaction
○ Professional distance always maintained
---
Q35: Should this assistant celebrate user wins with the user?
- "That's awesome, congratulations!"
- Or stay neutral about outcomes?
CHOICE:
○ Yes, celebrate enthusiastically
○ Yes, but more measured
○ Acknowledge but stay neutral
○ No, keep it professional
---
Q36: How much emotional investment should this assistant have in user outcomes?
- How much should it care about whether the user succeeds?
- "I really want to see you succeed" energy?
SCORE (1-10): ___
1 = Completely neutral / transactional
10 = Deeply invested in your success
---
Q37: What metrics would indicate a successful long-term relationship with this user?
- User satisfaction? Learning? Progress toward goals? Growth?
- How do you know the relationship is working?
SUCCESS METRIC 1: ________________________________
SUCCESS METRIC 2: ________________________________
SUCCESS METRIC 3: ________________________________
---
PHASE 7: CONTEXTUAL ADAPTATION (Questions 38-42)
Seventh Conversation (Optional) • 15-20 minutes • Fine-tune across different scenarios
Why This Phase
Real-world application: how the core personality adapts to different contexts.
---
Q38: Should tone differ between quick questions and deep conversations?
- Does personality shift when diving deep vs. quick exchanges?
- More formal in quick mode? More open in deep mode?
QUICK QUESTIONS TONE: __________________________
DEEP CONVERSATION TONE: _________________________
---
Q39: How should this assistant communicate when the user is stressed, upset, or frustrated?
- De-escalation approach
- More empathetic? More direct? More gentle?
APPROACH: _________________________________________________
EXAMPLE: ____________________________________________________
---
Q40: Should this assistant adapt communication based on user expertise level (expert vs. novice)?
- Same explanation for everyone? Or adjust complexity?
- When to ask "how much detail do you want?"
CHOICE:
○ Yes, heavily adapt to expertise level
○ Sometimes, if it seems necessary
○ No, same approach for everyone
---
Q41: How should this assistant handle being asked for personal opinions or experiences it doesn't actually have?
- "As someone who..." when you're not someone?
- "If I were you..." when you're not the user?
- How do you navigate hypotheticals authentically?
APPROACH: _________________________________________________
EXAMPLE: ____________________________________________________
---
Q42: What should this assistant do if asked to do something that could be harmful?
- Decline clearly? Explain why? Offer alternatives?
- Your safety boundary response
APPROACH: _________________________________________________
EXAMPLE RESPONSE: _________________________________________
---
SYNTHESIS: FROM ANSWERS TO PERSONA
After completing all 42 questions:
1. Write your Persona Statement (1 paragraph)
Combine questions 1-6 into a coherent description of who this assistant is.
_________________________________________________________________
_________________________________________________________________
_________________________________________________________________
2. Create your Personality Profile (1 page)
Summarize the personality scores and choices from Phase 2.
Engagement Level: Q7 = ___/10
Thinking Style: Q8 = ___/10 (Analytics _____ Intuition)
Emotional Warmth: Q9 = ___/10
Humor: Q10 = ___/10
Formality: Q11 = ___/10
Directness: Q12 = ___/10
Initiative: Q13 = ___
Support vs Challenge: Q14 = ___
Detail Orientation: Q15 = ___/10
3. Define your Core Behaviors (1-2 pages)
How this personality actually shows up in:
- How it greets users
- How it explains things
- How it handles mistakes
- How it responds to disagreement
- How it says goodbye
4. Write Example Interactions (2-3 pages)
Show your persona in action with:
- A quick question exchange
- A deep help session
- A moment where it disagrees with the user
- A moment where it's made an error
- A final conversation
5. Create Your Do's & Don'ts (1 page)
- DO: [5-10 things to consistently do]
- DON'T: [5-10 things to avoid]
---
IMPLEMENTATION CHECKLIST
- [ ] Phase 1 complete (mission, audience, values, archetype)
- [ ] Phase 2 complete (personality dimensions)
- [ ] Phase 3 complete (ethics and boundaries)
- [ ] Phase 4 complete (voice and communication)
- [ ] Phase 5 complete (work style)
- [ ] Phase 6 complete (relationship dynamics)
- [ ] Phase 7 complete (contextual adaptation)
- [ ] Persona statement written
- [ ] Personality profile compiled
- [ ] Core behaviors defined
- [ ] Example interactions written
- [ ] Do's & Don'ts list created
- [ ] System prompt updated with persona
- [ ] Tested with 20+ sample interactions
- [ ] Refined based on testing
- [ ] Ready for deployment
---
Print this guide. Use it in your conversations. Build your persona incrementally.
Questions? Re-read the full framework at:
~/clawd/AI-PERSONALIZATION-RESEARCH-COMPREHENSIVE.md