Inside Silicon Valley's $600B Startup School

How Vivek and John did it: Inside Silicon Valley's $600B Startup School

Vivek and John have started a lot of stuff together. They met freshman year of Penn, went their separate paths after graduating, and came back together because the desire to create and launch their own things was always there. The Y Combinator interview call landed while they were on vacation in Japan — they'd applied two months earlier and got two days' notice, which meant immediately ending vacation and going back into grind mode, working Pacific hours over a table in a Sapporo Airbnb to prepare. One of the people who interviewed them was literally the founder of Google Photos; Michael Sybel sat in on the call too. They applied with one idea the partners disliked a lot — but the partners liked another idea, and that other idea is what they've been working on since. As John puts it, turns out the professional investors were right about their old idea.

What got them in wasn't the idea. YC's view is that startups don't fail because of money — it's more so because of co-founder breakups. Vivek and John had known each other for seven years and built work together, and that relationship was able to drive past the fact that they had a bad idea. Inside the batch, the most beneficial part for one of them was the weekly dinners, where 100 to 200 founders ate together and speakers like the Airbnb and Doordash founders came to tell their stories. The cohort system put roughly ten companies in a little pod that met weekly to set metrics and push each other — helpful and somewhat competitive at the same time.

The lesson that reworked John's mind was sell before you build. The best thing is always to launch, see who's actually interested in your product, before you waste time building a whole pitch deck or a perfect MVP. The easiest way to figure out if you're solving a pain point is whether people pay for it, and how quickly. John's college mistake was sending out surveys and building waitlists to create demand — people fill out forms because they're your friend or just want to fill out a form, which is night and day from a customer paying $200 per seat. A big lesson they learned the hard way: do paid pilots, not free ones. One company led them on for a month or two, signed an NDA, said they were excited to build, then ghosted after four or five follow-ups — and still liked John's posts on LinkedIn.

Affil.ai is building AI compliance for financial affiliate marketing — making sure a content creator advertising the Chase Sapphire Preferred or the Amex Gold does it correctly, because if they don't, the banks could take back commissions or kick them out of the programs for what's technically false advertising. John already knew this pain point: he has 43 credit cards, does it for fun, became an affiliate himself back in 2021 or 2022, and later launched the entire affiliate program at a travel startup before this one. That compliance experience became the wedge into the space. The first customer arrived when they heard the pain point, told the customer they could solve all their compliance problems — and then John became the AI. "The AI is just me," he says: the customer would give him thousands of documents and he'd skim-read them at blitz breakneck speed, validating manually with the AI barely working, while showing slides and saying the AI worked great. They called it John Gpt. After a month of John reading garbage, the customer signed the contract — and that was the turning point where it was fully time to build.

On hiring, the rule was to wait until they started missing first-customer demands. With first and second customers that were pretty big enterprises and just the two of them, they hit a red flag when they actually missed something during a crucial moment of a contract — that's when they brought someone on. The qualities they looked for in engineering: being able to figure things out, experiment, iterate fast, and take ownership without being handheld. On money, the biggest mistake to avoid is being a perfectionist — they took too long to launch crafting a perfect story, and they had to work hundred-hour weeks to make their scrappy first deal work. Their advice: money is everywhere, only take it if it can drive a catalyst for growth, and focus on revenue and growth over profit.

What you'll hear

• The interview that disliked their idea — applying to YC with one idea the partners disliked a lot, while the founder of Google Photos and Michael Sybel grilled them
• Why the relationship beat the idea — startups fail from co-founder breakups, not money, and seven years of working together carried them past a bad idea
• Sell before you build — the YC mantra that reworked John's mind: launch first, see who pays, before wasting time on a pitch deck or perfect MVP
• Paid pilots, not free ones — the company that signed an NDA, said it was excited, then ghosted after four or five follow-ups
• John Gpt — how John manually skim-read thousands of documents at "blitz breakneck speed" to fake the AI for the first customer
• When to hire — waiting until you start missing your first customers' demands, after actually missing something during a crucial contract moment
• What Affil.ai actually does — AI compliance for financial affiliate marketing, so creators advertise the Chase Sapphire Preferred or Amex Gold without the banks clawing back commissions

Key claims from this episode

Chapters

Quotes from this episode

Themes Vivek returns to

• Sell before you build — launch first and see who actually pays, instead of wasting time on a pitch deck, a perfect story, or a perfect MVP
• Do things that don't scale — John manually validating thousands of documents as "John Gpt" because the first job is solving the pain point, not building the product
• Pay is the only real signal — surveys and waitlists tell you nothing; whether and how fast someone pays tells you everything
• Relationship over idea — seven years of working together is the thing that carried them, because co-founder breakups kill startups
• Don't be a perfectionist — they took too long to launch chasing a perfect story, and learned that solving the pain point matters more than a polished product
• Hire only when you're missing demands — wait until you actually start missing your first customers' needs before bringing anyone on