Gokul Rajaram - Lessons from Investing in 700 Companies - [Invest Like the Best, EP.456]

What It Covers

Gokul Rajaram shares lessons from building ads products at Google and Facebook and investing in 700 companies. He explains how AI transforms product development, the three ways ad businesses make money, sources of defensibility in AI-native companies, and what makes products durable when software becomes cheap to create but hard to defend.

Key Questions Answered

• •Product development transformation: Product managers now check code into production repositories using Claude or Codex, with AI soon reviewing code before engineers commit. The PM-to-engineer ratio shifts from one-to-ten to one-to-twenty as designers and PMs merge roles. PMs focus on articulating customer needs and owning evaluation systems rather than prescribing features, while AI handles design work within established design systems.
• •Three monetization models for ads: Ad businesses succeed through exactly three approaches: owning coveted first-party user inventory like Google Search or Facebook, driving specific outcomes at scale like Applovin does for mobile app installs, or serving as exclusive provider for large advertisers like Trade Desk does for Procter and Gamble. Companies attempting to be middlemen on top of Google or Facebook get squeezed as platforms incorporate their capabilities.
• •AI durability framework: Companies need scarce assets like licenses or regulations, control points over money or data flows, hardware components, essential workflow integration, or network effects to survive. Every AI-native company must plan to replace entire legacy systems of record, not just build workflow layers, because companies like Slack now block API access or charge prohibitively to protect their data moats.
• •Self-serve product mandate: Products must enable customers to onboard and use them without contacting any employee. Larry Page forced Google AdSense to give small customers the same advanced tools built for large enterprises, revealing that self-serve users exploit systems more creatively than sales-supported ones. Self-serve enables reaching millions versus thousands through sales teams and allows bottom-up infiltration even against incumbents.
• •Outcome-based customer behavior: Product managers must articulate every feature as a hypothesis about customer behavior change, stating specifically how users will shift from state X to state Y. North Star metrics should indicate both customer value and business growth, paired with check metrics as guardrails. At Facebook, the ads team received an annual engagement budget limiting how much user engagement could decrease in exchange for ad revenue.
• •Founder authenticity assessment: The founding story reveals whether entrepreneurs have authentic lived experience compelling them to solve a specific problem versus simply wanting to start a company with friends. Dylan Field of Figma was steeped in design thinking. Max Rhodes of FAIR solved distribution problems he faced running an undergraduate umbrella company. Founders must navigate the idea maze, explaining why they chose their solution over five or six alternative approaches.