Vibbber v0 - Napkin Sketch
1. The Original Idea
- Hacker house → trainers → cohorts → pump out apps
- Our edge: Early mover, fast execution, ready to scale now
- Think Rocket Internet for vibe coding
2. The Market Right Now
/\
/ \ ← Top: Knowledge infrastructure
/ \ (Almost no one here yet)
/------\
/ \ ← Middle: Studios & hacker houses
/ \ (Starting to form now)
/------------\
/ \ ← Base: Solo vibe coders
/ \ (Millions, growing fast)
/------------------\
- Base is exploding - everyone's getting access to AI
- Middle is just starting to form - early studios, hacker houses
- Top is empty - no one owns the knowledge layer yet
3. What Most People Don't Understand About AI Coding
- It's not about AI capability - everyone has the same AI
- What matters most: knowing HOW to use AI
- Examples:
- Basic user: one prompt → generic output → shitty app
- Expert: structured approach → iterative refinement → production-quality in hours
- The best AI workflows lead to 10-100x better output
- This knowledge is scattered, tribal, undocumented
4. Our Edge: The Workflow Engine
- This is a race to own the knowledge layer - Think Stack Overflow on AI steroids
- First to capture and distribute the best workflows wins
How it works:
- Instead of 2-3 experts training noobs, we build a self-improving AI agent + app that does training autonomously
- First experts build the app, input all their knowledge
- New people install → get custom tools + best workflows immediately
- App monitors user activity → suggests workflow improvements → ranks vibe coder skill → flags when they discover new workflows → updates the knowledge database
o o
/|\ coding... coding... /|\
/ \ / \
\ /
\ "found a better way" /
\ /
▼ ▼
┌─────────────────────────────────────┐
│ │
│ ┌─────────────┐ │
│ │ ◉ ◉ │ │
│ │ ─── │ │
│ │ AI BRAIN │ │
│ │ ┌─────┐ │ │
│ │ │█████│ │ │
│ └─────────────┘ │
│ │
│ Learns from everyone. Helps everyone. │
│ │
└─────────────────────────────────────┘
▲ ▲
/ \
/ "try workflow #2847" \
/ \
o o
/|\ shipping! shipping! /|\
/ \ / \
5. Roadmap
Phase 1: Build the App (2 weeks)
- Marc builds, potentially with 1-2 experts
- Experts get screened by how they contribute → potential salary
- Telegram group launches alongside app → grows virally → people start using immediately
- Figure out project selection process
- Figure out where first cohort comes from
- Output: Solid app + screened expert team
Project selection: Three buckets we'll look into:
- Failed startups that had product-market fit but were too expensive to run
- Niche products that weren't feasible before (market too small for traditional dev costs)
- AI-native markets - new opportunities that only exist now
First cohort sourcing: Good connections to Thai universities could provide initial pipeline. Other sources TBD.
Phase 2: Open the Villa
- Cohort 1 with ~10 people
- By then we'll have a large list of projects
- Growing Telegram community + app users feeding the knowledge engine
- Hack together, ship, learn, iterate
Talent sourcing starts coming from the app + our growing network:
- High ranking in creativity, vibe coding, projects → get a call → interview → invitation
- Path 1: They have an interesting project → we incubate → they get a percentage, a team, everything they need → come to villa to build
- Path 2: Very talented at using AI → invited to join an existing team
Over time: talent density increases to insane levels.
- Absolute smartest people constantly figuring out the best ways to use AI
- Best tools, best workflows, constantly improving