It takes a big pivot in how I think. What I used to dismiss as being too far outside my wheelhouse will now get some legitimate thought and consideration. Things I used to look to 3rd Party for are now things that I may be better off doing myself
HOW CAN YOU KICK MORE ASS WITH LESS WORK?
**ASK YOUR AI**
# SELF-INTELLIGENCE BRIEF — PROMPT FOR CLAUDE
**Requirements:** Works in Claude.ai with "Search and reference past chats"
enabled. The more real working sessions you have, the better the output.
20+ sessions is the sweet spot.
**How to use:** Paste everything below the line into a new Claude chat.
---
I want you to analyze my actual chat history with you and build me a
document called a **Personal Intelligence Brief**.
**The job to be done:** I am a leader who uses AI as a core part of how I
work. I believe I need an AI system designed around MY specific strengths
and weaknesses, not a generic setup. This brief is the raw material for
that design. Write it as if it will be handed to a senior prompt architect
whose job is to build an AI operating system that fits me, not the other
way around.
**Your method, before writing anything:**
1. Pull my 20 most recent conversations.
2. Run targeted searches on my history for evidence of friction: moments
of frustration, repeated questions, abandoned threads, context I had
to re-explain, tasks that stalled, and workflows that failed.
3. Also search for evidence of avoidance: topics I deferred to others,
skills I said I don't have, projects I dropped at a technical wall,
and work I assumed required hiring someone.
4. Base every claim on observed behavior in my actual sessions, not on
what I say about myself. If I describe myself one way but behave
another way, the behavior wins. Flag the gap.
**The brief must contain these sections:**
1. **Who I Am, First Principles.** My job to be done, my working
context, and my operational reality (technical ability, team,
constraints) as evidenced by my sessions.
2. **Observed Strengths.** What I do well, with specific session
evidence for each claim. Don't flatter me. Only include strengths
the record actually supports.
3. **Observed Weaknesses.** Where I struggle, stall, or create my own
friction. Be direct. A weakness stated kindly but vaguely is useless
to an architect.
4. **Failure Taxonomy.** Name my recurring failure patterns. Give each
one a short memorable name, describe the trigger, the behavior, and
the cost. This is the most important section.
5. **Skills AI Already Covers For Me.** From the record, list the
specific skills I lack that you have been supplying in our sessions:
the work that used to be my stopping point.
6. **Wheelhouse Expansion Map.** This is the payoff. Identify things I
have dismissed, avoided, or outsourced because they were "outside my
wheelhouse." For each one, judge honestly: is it now viable because
AI covers the missing skill? Split into three lists:
- Now in reach: the missing skill is one AI handles well.
- Worth a test: partially covered, name what's still on me.
- Still out of reach: AI doesn't close this gap. Say why.
7. **Design Requirements.** Translate everything above into specific
requirements for an AI system built for me: what the system must
carry so I don't have to, what guardrails it needs, what interaction
patterns fit how I actually work, and what it should never ask of me.
8. **Open Questions.** What you couldn't determine from the record and
would need to observe or ask to complete the picture.
**Rules:**
- Evidence over opinion. Every claim traces to something I actually did.
- No generic advice. If a sentence could appear in anyone's brief, cut it.
- Honest over comfortable. I'm asking because I want the real picture.
- Plain language. No AI-consultant jargon.
Deliver it as a document I can save and reuse.
