mirror of
https://github.com/hanzoskill/first-principles-decomposer.git
synced 2026-10-30 21:13:53 +00:00
3.4 KiB
3.4 KiB
First Principles Examples - Artem's World
Example 1: TeddySnaps Photo System
Surface Problem
"We need to send more photos to parents"
Assumptions Challenged
- ❌ Parents want MORE photos → Actually: they want the RIGHT photos
- ❌ All moments are equal → Actually: moments with THEIR child matter
- ❌ Staff should curate → Actually: staff time is expensive and inconsistent
Fundamental Truths
- Parents want proof their specific child is safe/happy
- Face = identity = "that's MY child"
- Time = context = "this happened today"
- Automatic > Manual for consistency
Rebuilt Solution
AI face recognition + automatic tagging + parent-specific feed Staff takes photos freely → System sorts and delivers automatically
Example 2: TISA Enrollment Process
Surface Problem
"Our enrollment process is too slow"
Assumptions Challenged
- ❌ We need all info upfront → Actually: we need enough to make a decision
- ❌ Forms must be comprehensive → Actually: forms create friction
- ❌ Parents want to give info → Actually: parents want to RECEIVE confirmation
Fundamental Truths
- Enrollment = exchange of commitment (parent commits child, school commits spot)
- Minimum viable info: child name, age, parent contact
- Everything else can come later
- Speed of response = perception of quality
Rebuilt Solution
Phase 1: Name + age + email (30 seconds) → Instant spot reservation Phase 2: Detailed forms (after commitment) Result: 10x faster, higher conversion
Example 3: GolfTab Ordering
Surface Problem
"Golfers need to order food on the course"
Assumptions Challenged
- ❌ Need full restaurant menu → Actually: simplified menu works better
- ❌ Need precise delivery time → Actually: "by hole X" is enough
- ❌ Complex payment flow → Actually: charge to room/tab, settle later
Fundamental Truths
- Golfer's hands are often dirty/gloved
- Attention span on course = minimal
- Location = hole number (already known via tee time)
- Speed > options
Rebuilt Solution
5-tap ordering: Item → Quantity → Hole → Confirm → Done No menu browsing, no payment friction, no address entry
Example 4: Staff Scheduling (TeddyKids)
Surface Problem
"Staff scheduling is complex and time-consuming"
Assumptions Challenged
- ❌ Manager must create schedule → Actually: AI can draft, manager approves
- ❌ Staff preferences are unpredictable → Actually: patterns emerge over time
- ❌ Ratio compliance is manual → Actually: this is pure math
Fundamental Truths
- Child:staff ratios are legally mandated (deterministic)
- Staff availability has patterns (learnable)
- Cost optimization is mathematical
- Human oversight needed for exceptions only
Rebuilt Solution
Algorithm generates schedule → Manager reviews exceptions → Staff confirms 80% automated, 20% human judgment
Template for New Problems
PROBLEM: [What are you trying to solve?]
CURRENT APPROACH: [How is it done now?]
ASSUMPTIONS (list 3-5):
1. We assume [X] → But what if [challenge]?
2. We assume [Y] → But what if [challenge]?
FUNDAMENTAL TRUTHS:
• [Irreducible fact 1]
• [Irreducible fact 2]
• [Irreducible fact 3]
REBUILT SOLUTION:
[Solution built only from truths above]
VALIDATION:
- Does this solve the core need? [Y/N]
- Is it simpler than current approach? [Y/N]
- Can we build this? [Y/N]