Skip to content
Y CombinatorY Combinator

How Internships Seed Most of YC's Billion-Dollar Startups

Internships at Cohere and going undercover as a medical biller surface real problems; hackathon ideas miss these, but outsourced jobs signal the next wave.

Garry TanhostHarj TaggarhostJared FriedmanhostDiana Huhost
Feb 6, 202543mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

From Hackathon Hacks To Billion-Dollar AI: Finding Real Ideas

  1. The hosts of the Lightcone Podcast explain how aspiring founders can systematically generate strong AI startup ideas by going beyond shallow, hackathon-style projects and instead pursuing hard, ambitious problems.
  2. They emphasize two main paths: looking deeply within your own past work, internships, and skills for unique founder–market fit, and aggressively going outside your comfort zone to embed yourself in real-world industries and workflows.
  3. Through many concrete YC portfolio examples, they show how founders uncovered valuable ideas by leveraging specialized experience, following cutting-edge technology, shadowing or working in tedious jobs, and serving overlooked niches now made viable by AI.
  4. They argue that despite apparent competition and investor skepticism, this is an unusually good time to persist in AI startups, because capabilities and opportunities are expanding so quickly.

IDEAS WORTH REMEMBERING

5 ideas

Start from genuine expertise or experience, not from what’s trendy.

Many of the strongest AI ideas came from founders who had spent years at the edge of a domain (e.g., Tesla finance, Figma QA, Cohere, Scale, hardware design) and saw problems only they were well-positioned to solve.

If you lack expertise, deliberately go get it in the real world.

Founders created great companies by shadowing relatives, working a day in a dentist’s office, getting a medical billing job, or closely observing friends’ boring jobs, then using AI to automate the painful workflows they saw.

Treat early exploration like research, not short-term revenue chasing.

When you don’t yet have a strong idea, the hosts suggest acting like a researcher or PhD: go to the edge of some field, build understanding and prototypes, and let insights about real problems emerge over time.

Aim for harder, more ambitious ideas that truly capture imagination.

Founders tend to gravitate to what can be built in a weekend; the podcast urges them to push toward ideas that are technically hard, have big upside (e.g., universal translators, new social networks), and would excite them for years.

Look for jobs and workflows that are outsourced, repetitive, or consultant-heavy.

Tasks already offshored or requiring armies of consultants (drive-thru order-taking, RPA deployments, government-bid monitoring, back-office insurance processing) are strong signals that AI can create high-value automation startups.

WORDS WORTH SAVING

5 quotes

There is a very important question that all founders have to ask when they commit to an idea, and that is: if not us, then who?

Garry

Often when people are thinking about startup ideas, they tend to gravitate towards things that seem very easy to build a first version of… most of the best startup ideas are actually at least somewhat hard to ship the first version of.

Jared

Work on something that captures the human imagination.

Chris, as quoted by Jared

If you’re building a startup working on cutting-edge AI, even if you haven’t found the right idea yet, why give up and go back to Google or college or something?

Harj

You can’t stay in your house or sit on your toilet doomscrolling X. You have to either look very deeply within you… or you need to radically get out of the house.

Garry

Avoiding shallow, hackathon-style AI startup ideasLeveraging founder expertise and internships for founder–market fitGoing “outside the house” to study real-world industries and workflowsUsing AI to automate manual, outsourced, or admin-heavy knowledge workFinding ideas via family, friends, and undercover or shadow jobsLiving at the technological edge and noticing what’s missingDealing with competition, pivots, and investor skepticism in AI startups

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome