Building real things: how to thrive in tech during the AI era

Executive overview

CS graduates now face higher unemployment than art history majors, while tiny AI-native teams reach eight-figure revenue in months. The old path — CS degree, stable job, credential ladder — is no longer low-risk.

The opportunity isn't closing; it's inverting. Technical depth combined with agency to go learn a niche domain is the new edge, and college students are often better positioned than veterans to exploit it.

The credential trap and why it's broken

• CS majors hit 6.1% unemployment in early 2025 vs. 3.0% for art history majors.
• The "safe" FAANG job was always credentialing obedience and reliability — skills AI now does better.
• Raising a VC round is a credential, not an outcome; revenue is the outcome.
• Entrepreneurship programs that teach startup as a checklist are producing facsimiles, not founders.
• Two dangerous self-imposed credentials: optimising for investor approval; treating a fundraise as the goal.

Why this is the best time to build

• A few people are going from zero to $10–12M net revenue in 12 months — unprecedented for B2B SaaS.
• Cursor went zero to $1B in year one, then one to $100B the next — an order of magnitude shift in what's achievable young.
• AI has captured every domain's imagination; dentists, logistics operators, and specialists are actively open to "magic in a bottle" from college students.
• The wedge available now is not software features — it's AI doing the actual work of people, which commands far higher prices.

The domain expertise gap — and how to close it

• Pre-AI, domain expertise (relationships, sector knowledge) mattered more than technical skill; markets were saturated.
• Post-AI, the technical edge is back and is now the missing piece most incumbents can't supply.
• College students consistently outperform PhDs on squeezing reliable performance from models.
• You can go from zero domain knowledge to genuine expertise in a month or two with focused effort — most people underestimate this.
• The method: become a forward-deployed engineer — go undercover, embed in a business, watch how the work actually happens, then build.

Niche is the strategy

• Every major platform company started absurdly niche: Airbnb (air beds at Democratic conferences), Stripe (instant API payments for weekend projects), Coinbase (regular people buying Bitcoin with a clean UI).
• AI makes niche more valuable, not less — because you're selling work, not software, so willingness to pay is far higher.
• The more obscure the niche to outsiders, the more durable the moat.
• Find 10 users who love the product obsessively; ignore 100 lukewarm ones.

Agency over credentialism

• Most students are conditioned to pass tests in constrained boxes; startups have no predetermined rules.
• "What should I look like to raise money?" is a student question. There are no adults in the room — you set the rules.
• "Is this the last window to get rich?" is a fear-framing that misses the point; if AGI makes the question real, money itself changes meaning.
• Build from excitement about what you're making, not fear of missing the window.

On social media and building in public

• Aura-farming and flashy launch videos are simulacrum — they represent something but contain nothing.
• The real version: two-week sprints working backwards from a simple Loom video showing a concrete feat of strength.
• Storyboard what you want to demonstrate, build exactly that, share it — substance over flash.
• Own your own story; if you don't tell it, others will, and they'll prefer the unbecoming narrative.

Practical decisions: dropping out, joining a startup, quitting a job

• Drop out if: you trust the team, it's genuinely a strong startup, and you're actually done with what college has to offer you — not out of FOMO.
• Join a startup — but treat it like an investment: make a spreadsheet, evaluate it the way a VC would, then remember the VC has a portfolio and you have one life.
• To quit a full-time job: have six to nine months of ramen-level runway saved before you leave.
• First-time founders: don't go solo. The learning gradient is too steep; you need a co-founder. Co-founder timing is the biggest real bottleneck — if it aligns, just do it.
• Power law for startups is brutal: aim for superlative teams at superlative places, not median outcomes.