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 August 31, 2026
AI drives consolidation in regulated and incumbent industries
Pressure from increasingly capable AI systems is expected to trigger massive consolidation in banking, healthcare, and other regulated sectors, as only firms with sufficient scale can build defensible positions against intelligent automation.
- Keiretsu-style aggregation and compute hoarding emerge as defensive strategies.
- Incumbents with deep proprietary data have a durable training advantage.
- Few founders or executives have seriously modeled what this means for firm structure.
AI pressure could trigger massive consolidation across industries · @hypersoren AI may force massive consolidation in banking and healthcare · @WillManidis Thomson Reuters launches LLM trained on 175 years of proprietary data · @zachmoskow Incumbents can compound AI advantage by training on proprietary data · zachmoskow
Consumer AI agents cross a real-world utility threshold
AI assistants are shifting from novelty to genuine utility by handling live, context-aware tasks like rebooking reservations from conversational shorthand. The gap is closing between asking an agent to do something and having it actually done.
- Proactive scheduling and memory of prior context are key differentiators.
- Most businesses lack the infrastructure to receive agent-driven interactions.
- Multiplayer coordination between agents could reshape household and social logistics.
Five UX decisions behind Instinct AI assistant's product-market fit · @kushalbyatnal Lessons learned from launching an iMessage AI agent a year too early · @slopdotwtf Consumer AI agents now handle real-world bookings conversationally · @jeff_weinstein AI agent books and reschedules reservations autonomously · @sytaylor AI agents in multiplayer mode could replace nagging between partners · @scottbelsky Consumer agents need multiplayer mode to manage shared household tasks · Scott Belsky AI agents feel real when they remember context and change live bookings · jeff_weinstein Consumer agents get real when they can find, change and confirm bookings · @sytaylor
Founders trapped between early exits and permanent private status
Ambitious founders face a narrowing set of outcomes: sell quickly to an incumbent or stay private indefinitely as a vehicle for capital allocation. Neither path serves the goal of building an enduring, independently scaled company.
- Public markets remain an underexplored third path for private-capital-built companies.
- Venture market bifurcates between crowded consensus bets and ignored contrarian ones.
- Quick flips are profitable but foreclose the founder's original ambition.
AI compresses small wins but not category-defining scale
Building a $10M ARR software company is dramatically easier with AI, but reaching $1B remains as hard as ever. The ease of entry raises the floor while leaving the ceiling unchanged, which will define which startups and VC bets actually matter.
- Free software as a customer acquisition layer becomes viable as build costs fall.
- The $10M-to-$1B gap is where most venture returns will be won or lost.
- AI-native incumbents training on proprietary data may compound advantages at scale.
AI lowers the bar to $10M ARR but not to $1B · @grinich Free software as customer acquisition is becoming a viable model · @GJarrosson Thomson Reuters launches LLM trained on 175 years of proprietary data · @zachmoskow Incumbents can compound AI advantage by training on proprietary data · zachmoskow