Vibe Code Camp - Live Marathon
Date: 2025 Host: Every (Dan Shipper) Duration: 8 hours Source: YouTube
An all-day marathon featuring the world’s best AI builders showing their workflows, tools, and philosophies for building with AI agents.
Sessions
Keynote & Every Team
- Dan Shipper - Every CEO demos Proof (agent-native markdown editor) and Anecdote (health app)
- Katie Parrott - Using Claude as a writing partner, allocation economy concept
- Naveen Naidu - Building Monologue, solo dev competing with VC-backed voice apps
Non-Technical Builders
- Ben Tossell - Can’t code, ships production software with Droid. Session sharing, debugging mindset
- Tina He - “Tokens don’t lie” - vibe coding as meditation, building for joy
Production AI Development
- Ryan Carson - Infinite loop agent with AMP, compound product methodology
- Nat Eliason - Clawdbot: Mac Mini running 24/7 unsupervised, Conductor GUI
- Paula Dozsa - 200K MAU iOS app with zero iOS experience, Portola agency
Tools & Frameworks
- CJ Hess - Context7 MCP that teaches Claude any framework on demand
- Kevin Rose & Kieran Klaassen - Compound Engineering: plan, work, review, compound
- Ashe Magalhaes - Personal AI suite (ash.ai), multi-agent architecture
Enterprise & Research
- Natalia & Nityesh - 20-hour consulting decks in 2-hour workflows
- Brooker Belcourt - 5-hour hedge fund research in 20 minutes
- Yash Poojary - Reverse engineering ChatGPT and Spotify to learn faster
Platform Perspectives
- Logan Kilpatrick & Ammaar Reshi - Google AI Studio capabilities and roadmap
- Thariq Shihipar - Inside Anthropic: Claude Code philosophy, tasks replacing todos
- Geoffrey Litt - Malleable software vision, Notion-based agent management
Key Themes
The Model Does Most of the Work
Multiple speakers emphasized that complex setups (MCPs, hooks, skills) aren’t necessary. Simple prompts + good models = results.
Pain Threshold > Technical Skill
The barrier to vibe coding isn’t coding ability - it’s willingness to work through debugging loops and frustration.
Agent-Native Architecture
Building systems where agents have the same capabilities as users. If a human can do it, an agent should be able to do it.
Session Sharing as Learning
Exporting and reviewing complete agent sessions teaches patterns organically - better than tutorials.
Planning Before Coding
Spending time on specs and plans with the AI before implementation dramatically reduces debugging time.
Tools Mentioned
| Tool | Use Case | Mentioned By |
|---|---|---|
| Claude Code | Primary coding agent | Everyone |
| Droid (Factory) | Multi-model agent with compaction | Ben Tossell |
| Conductor | GUI wrapper for Claude Code | Nat Eliason |
| Compound Engineering | Claude Code plugin | Kevin, Kieran, Katie |
| Context7 | Framework docs MCP | CJ Hess |
| Monologue | Voice-to-text | Naveen Naidu |
| Google AI Studio | Prototyping | Logan, Ammaar |
| Supabase | Backend | Multiple |
| Vercel Agent Browser | Automated testing | Ben Tossell |
All Sessions (Chronological)
- Dan Shipper - Every, Proof, agent-native architecture
- Ben Tossell - Non-technical vibe coding, Droid, debugging
- Ashe Magalhaes - Personal AI suite, ash.ai
- Ryan Carson - Infinite loop agent, AMP, compound product
- Natalia & Nityesh - 20-hour decks to 2-hour workflows
- Katie Parrott - Writing partner, allocation economy
- Nat Eliason - Clawdbot, 24/7 Mac Mini agent
- Tina He - Vibe coding as meditation
- Paula Dozsa - 200K MAU app, zero iOS experience
- CJ Hess - Context7 MCP, teaching Claude frameworks
- Logan & Ammaar - Google AI Studios
- Geoffrey Litt - Malleable software, Notion workflows
- Kevin Rose & Kieran Klaassen - Compound Engineering philosophy
- Thariq Shihipar - Inside Anthropic, tasks replacing todos
- Naveen Naidu - Monologue, solo dev vs VC-backed
- Yash Poojary - Reverse engineering ChatGPT/Spotify
- Brooker Belcourt - Hedge fund research in 20 minutes