Garry Tan: New Rules for Founders

Founder Psychology & Career Trajectory

• 2003 Context: The speaker entered tech post-Web 1.0 crash, noting a bleak job market where the only options were Microsoft (Windows Mobile) or Expedia; he initially chose Microsoft due to student loan pressures despite a preference for startups.
• The "Hot vs. Known" Mistake: A primary career error was chasing perceived trends (mobile) instead of his established expertise in web programming, missing the emergence of Facebook and social software.
• Earnestness as Maturity: The speaker redefines "earnestness" not as naivety, but as the courage to trust one's direct experience and first-principles knowledge over external consensus or "map-reading" from investors.
• Palantir Opportunity Missed: The speaker declined a $70,000 equity check from Peter Thiel to join Palantir because he prioritized status (Microsoft Level 6 promotion) over the "smartest people," a decision he estimates as a $2–4 billion opportunity cost.
• Cultural Shift in Status: Working at companies like Palantir shifted from being low-status or viewed as a "services business" to being recognized as a critical, high-status entity with a compelling CEO.

Silicon Valley Culture & Access

• The "LARPing" Phenomenon: A current undercurrent in Silicon Valley involves "LARPing" (Live Action Role Playing) where founders follow trends rather than building genuinely useful products, often leading to the "permanent underclass" of those outside the light cone.
• Finding Your Tribe: The internet allows "outsiders" to instantly find their niche communities on platforms like Reddit or X, removing the historical necessity of physical presence in specific hubs to find peers.
• YC as an Equalizer: Y Combinator democratized access by replacing opaque social networks and school gatekeeping with a simple application process, treating ideas and execution as the primary metrics regardless of founder background.
• The "Cult" Dynamic: The speaker observes that "everything awesome" often starts as a cult-like belief system that contradicts orthodoxy, emphasizing that successful innovation requires a group of outsiders operating on a shared truth.

AI, Agents, & The Future of Work

• Productivity Multiplier: With AI coding agents, a single founder can now function as "400 people," enabling rapid execution where previously multiple engineers were required.
• "Markdown Files as Employees": The speaker proposes a new operational model where business processes are codified into reusable markdown files and agents, allowing for "feet of strength" that become permanent, cost-effective "employees."
• Token Maxing for 2028: To achieve full agent capability, founders must "token max" (load millions of tokens into context), a practice currently costing $50k–$100k/year but enabling agents to operate with "150 IQ" and access to deep institutional memory.
• Zero-to-$15M ARR: Companies are now capable of scaling from zero to $15 million ARR in four months with small teams (2–3 humans) supported by hundreds of agents and skill files.
• Conflict as Truth-Seeking: AI systems allow organizations to explore conflicting approaches without human emotional bias, turning "constructive conflict" into a mechanism for searching for truth rather than a source of drama.
• Provenance & Error Correction: As agent systems scale, "provenance" (tracking data sources) and automated conflict management become critical to prevent compounding errors in large-scale data environments.
• The "White Pill" of Slowness: The widespread adoption of AI will be slower than predicted because legacy institutions (government, Mag 7) are built on human cognitive limits (7±2 items), creating structural moats that will persist for decades.
• Agent-Native Founders: The next generation of founders (18–22 years old) are "AI-native," potentially outperforming traditional management structures by leveraging agents to bypass bureaucratic bottlenecks.
• API Line Erasure: AI erases the traditional "API line" where humans follow instructions; instead, agents can engage in back-and-forth negotiation with software, similar to the "Toyota Production System" where workers with the most context improve their own processes.

Organizational Design & Management

• Mid-Level Bureaucracy Replacement: The speaker argues that mid-level management layers should be replaced by agents that resolve conflicts and organize work programmatically, allowing executives to set direction while agents handle execution.
• Brex's "Crap Trap": Brex uses an open-source agent layer to audit network traffic and meeting transcripts, enabling the CEO to identify internal conflicts and operational breaks without being present in every meeting.
• The "Seven Plus Minus Two" Limit: Human cognition can only hold ~7 items at once, whereas agents can process entire organizations' data, enabling "clairvoyance" and management at a scale previously impossible.
• Microsoft Anecdote: A story from the speaker's time at Microsoft illustrates bureaucratic inefficiency: a P3 bug fix required a physical visit with a baseball bat as a symbolic threat because internal channels and emails were ignored by the Windows team.

Market Economics & Future Form Factors

• Consumer AI Economics: The speaker notes that high token costs currently prevent free-consumer AI models; however, a predicted drop in compute costs (to $50–$100/month) in 2–3 years will unlock a new wave of consumer software and "harness wars."
• Next Computer Form Factor: The future computer will likely prioritize voice interaction and persistent memory (ingestion, diarization) to become a benevolent "concern" that knows a user's hopes and fears.
• SaaS Viability: Pure "per-seat" SaaS models are becoming obsolete; successful future software must leverage data moats or network effects, as agents can replicate standard software functions.

Civic Engagement & Local Politics

• San Francisco Reversal: The speaker credits organized community action (specifically Asian American organizing around crime and education issues) with reversing the trajectory of San Francisco, which had become hostile to tech workers and elders.
• The "White Pill" of Rule of Law: The speaker's optimism stems from witnessing the rule of law and government earnestness in action (e.g., jury duty), countering the narrative that institutions are purely corrupt.
• Localism as Strategy: The speaker advocates for "act local" activism, arguing that solving hyper-local issues (housing, crime, education) creates functional systems that eventually influence state and national policy.
• Media Accountability: A key driver of local improvement was holding mainstream media accountable for ignoring anti-Asian crime and policy failures, forcing a re-evaluation of what "virtue signaling" actually looked like.
• Utopian Ambition: While acknowledging a perfect utopia is unattainable, the speaker maintains that attempting to build a better world through technology and civic engagement is a worthy and necessary endeavor.