From engineer to founder: conviction, culture, and AI employees

Executive overview

Repetitive enterprise work consumes talented people — even at the best companies. Surojit Chatterjee built Google's mobile ads business from zero to $50B by breaking internal norms and backing ideas others called impossible. He now applies the same conviction-driven approach to Ema, an AI company building universal AI employees for enterprise.

The willingness to risk being fired — for the right idea — is what separates incremental work from breakthrough products.

Building against internal resistance

• Google's smart-people consensus is a signal, not a veto — vehement opposition often means an idea has real potential
• Click-to-call ads faced massive internal resistance; it became a multi-billion dollar business
• Mobile features required nine months of data under Google's desktop-era experimentation norms — too slow for a fast-moving market
• Solution: deliberately broke the rule, launched to 50% of users, accepted the consequences, kept the feature when it worked
• Conviction means taking an idea all the way to a real test, not just defending it in meetings

Leaving comfort for learning

• Left a fast-growing, well-compensated role at Google specifically because comfort was becoming a trap
• Flipkart opportunity was serendipitous — Indian customers had already gone mobile-first, skipping desktop entirely
• India's e-commerce infrastructure required solving logistics from scratch: custom narrow trucks for small-town alleys, cash-on-delivery for the unbanked
• Each move — Google to Flipkart, back to Google, then Coinbase — was chosen for the steepest learning curve, not the safest bet

Culture degenerates without deliberate defence

• Growth always risks cultural drift — new hires, new norms, new distractions
• Google's open-discussion culture became a liability when it drifted outside the company's core mission
• Running a company is not a popularity contest — trying to please everyone guarantees culture decay
• Culture must be actively defended: name it, enforce it, accept being unpopular when someone doesn't fit

Building Ema: focus and stealth

• Worked with design partners from day one to validate demand before building broadly
• Critical distinction: identify which parts of the product are a reusable platform vs. what's being customised for early customers
• Kept the company in stealth for a year — no website, no fundraise announcement — to stay focused
• Noise is abundant; conviction and focus are the scarce resource
• Long-term vision: a conversational operating system for every enterprise, where small human teams work alongside AI employees to build billion-dollar companies