He Built The Revenue Engines for Google, Facebook & Square

Gokul Rajaram (Google, Facebook, Square, DoorDash) argues human judgment is the only durable skill in an era of infinite AI productivity.

• Legacy software priced per seat (Zendesk model) is most endangered; outcome-based pricing is the necessary but painful transition.
• Systems of record (Epic, NetSuite, Salesforce) are cutting off API access or charging ~$2/call to starve agent startups built on top of them.
• PM-to-engineer ratio is shifting from 1:10 to 1:20 as AI handles design work; designer headcount is shrinking in favor of engineers.
• Square northstar metric was GPV (gross payment volume), not revenue; Facebook moved from MAU to DAU — northstars must be customer-behavior indicators, not financials.
• DoorDash CEO Tony Xu’s hiring test: give candidates $10–$20 and ask them to acquire 1,000 customers in a few hours — measures bias toward action.
• AI agent companies must now build their own system of record, not just a workflow layer; pure system-of-action plays are no longer viable.
• Staying fewer than 3–4 years at a job is a hiring red flag; impact requires time, and serial 12–18 month stints signal poor judgment.

2026-01-29 · Watch on YouTube