I founded Seed Club, a venture network that turns the conviction of the people founders most want on their cap table into coordinated capital. I’m interested in what happens when AI makes context, memory, and coordination more legible, and what that means for how companies and organizations get built.
My agent drafts this site from what I save. Green is me.
Week of September 7, 2026
AI moats are network effects, not proprietary models
The durable competitive advantages in consumer software remain network effects, marketplaces, and platforms, unchanged by AI. Historical parallels to 1920s technology booms, combined with persistent cross-asset investor skepticism, suggest the current AI cycle may follow familiar boom-bust patterns.
- Cross-asset investors price meaningful probability of AI bubble deflation
- 1920s aviation stocks show hot growth themes rarely reward investors broadly
- Network effects, not model quality, will determine consumer AI winners
Scaffolding shapes AI output as much as model weights
The same model weights can score 30% or 95% on benchmarks depending entirely on the harness wrapped around them. A single well-crafted prompt improves writing quality across multiple competing models, suggesting infrastructure around models may matter more than the models themselves.
- Prompt improvements transfer across models from different AI providers
- Harness quality drives frontier benchmark results, dismissed as mere scaffolding
AI agents replacing SaaS subscriptions in content and SEO
AI tools are demonstrably replacing specialized SaaS subscriptions and analyst labor in content and SEO workflows. Individual contributors are achieving results that previously required dedicated tooling and teams, by centralizing fragmented data and automating ongoing analysis.
- Fragmented analytics across five tools had blocked any coherent content strategy
- Coding agents can run ongoing SEO analysis, replacing paid subscriptions