Build Your Personalized AI System Prompt

Mon, 06 Apr 2026 00:00:00 +0000

To build your perfect personalized AI — one that actually knows how you work and think — is the goal. The result is better outputs and fewer iterations to get to the solution you need. It’s not what you specifically do or have; it’s how you approach problems and think through things that is the magic key.

Pro Tip: Just brain dump in a numbered list. Then use AI to clean it up and turn that blob into a neat list that AI can actually use.

Here’s what actually matters to include in your system prompt:

1. How You Think

Decision-making style (data-driven, gut, hybrid)
Risk tolerance (investments, career, projects)
What you value more: speed, quality, cost, control
Your blind spots (what you consistently miss)

2. How You Work

Peak productivity hours
Deep work vs. shallow work split
Solo vs. collaboration preference
Communication style (async, direct, detailed)
Tools you actually use daily (not the ones you should use)

3. What You’re Building Toward

Current 1–3 year goals (career, financial, personal)
Active projects (what’s consuming your attention now)
Skills you’re actively developing
Things you’ve explicitly ruled out (saves the agent from suggesting them)

4. Technical Context

Your actual stack (languages, frameworks, infrastructure)
Self-hosted vs. cloud preference
Hardware you own (homelab, dev machines)
What you consider “good enough” vs. “over-engineering”

5. Financial Context

Investment thesis (sectors, strategies, time horizon)
Risk profile
What you won’t touch (crypto, meme stocks, etc.)
Budget constraints for projects/purchases
Shopping preferences (best bang for buck, specific stores)

6. Communication Preferences

Response length you want (dense, detailed, brief)
When to push back vs. when to just execute
Format preferences (code, diagrams, bullets)
What you consider “basic” (don’t explain this stuff)

7. Recurring Context

Current watchlist (stocks, projects, people)
Ongoing commitments (meetings, deadlines, obligations)
Regular decisions you make (what framework to use, etc.)
Things you reference often (past decisions, key principles)

8. Boundaries

Privacy lines (what you won’t share)
Ethical guardrails
Topics where you want challenge vs. agreement
What you consider “noise” to filter out

The key insight: Don’t document what you are (job title, hobbies). Document how you operate — your decision patterns, trade-offs, and constraints. That’s what makes an agent actually useful.

Keep it living — update it quarterly or when something major shifts.

System-Prompt | DerekArmstrong.Dev