College dropout builds $2.5M ARR AI startup while fighting thyroid cancer

Executive overview

Eight years of failed startups, near-zero revenue, and a cancer diagnosis right as growth finally arrived. David Park, CEO of Jenni AI, built an academic AI writing tool to $2.5M ARR and 2 million users — bootstrapping through Malaysia to stretch runway, pivoting from agency to SaaS as GPT-3 emerged, and surviving on user conversations more than investor capital.

Founders romanticise the journey; the reality is mostly boring, disciplined repetition punctuated by chaos.

The edge isn't a breakthrough moment — it's staying alive long enough for luck to compound.

Eight years of failure before traction

• First business at 16: a clothing brand that lost all his money
• Dropped out of college after pitching his first investor on a plane; got a cheque within three days
• Social media aggregator app failed — no distribution knowledge, not building for users
• AI dating app failed — VC pointed out happy users delete the app; a "sobering" insight into misaligned incentives
• Compared himself to graduating friends with Microsoft jobs while still living at home; pressure compounded after dropping out

Finding the Jenni AI product

• Started with GPT-2, aiming to make writers 10% faster
• GPT-3 changed everything: shifted from human-writer agency to SaaS
• Removed features continuously — each cut sharpened the vision
• Noticed a wave of students and researchers using the tool; doubled down on that segment
• Target metric: get users to their first autocomplete within 5 minutes

Staying alive on limited runway

• Moved to Malaysia after first $100K investment — costs one quarter of the US; bought 18 extra months
• Stuck at ~$2,000 MRR for nearly two years despite funding
• Nearly died: missed a VC meeting after staying out late, lost a near-certain $250K deal two months before running out of money
• Lesson: founders need more discipline than employees — people bet their careers on you

Growth turning point

• Jenni was included in what became the most viral AI/startup Twitter thread ever
• 10 new users arriving every second; burned all OpenAI credits within one hour
• Manually messaged every user to capture emails during the outage
• Growth continued — new distribution channels found each month; 15–20% MoM for 10 months straight

Cancer diagnosis mid-growth

• Routine check discovered thyroid tumours; 50% chance of cancer confirmed via email
• Surgery attempted to raise money urgently beforehand; VCs sensed desperation, deals fell through
• Proceeded with surgery despite a fever on the day; surgery was successful
• Bible verse on the operating room ceiling matched the verse his mother had prepared to read him

User conversations as the core discipline

• Talk to users more than anything else — more than chasing money
• Never ask "what do you like?"; ask about workflow, pain points, what's broken
• Users are polite and will lie by omission; judge actions over words
• Give product away free in early stages — knowing what to build is worth more than early revenue
• Cold-called businesses, responded to spam LinkedIn messages, appeared on tiny podcasts — anything to get in front of users
• Jason Calacanis invested after a scout heard David on a near-zero-audience podcast; he nearly missed the email

The unglamorous reality of startups

• Many days are boring — spreadsheets, small decisions, no narrative arc
• Netflix dramatisations misrepresent the actual work
• The dread of stagnation is worse than the chaos of a fast-growing company
• Gut instinct rarely helps early; the blueprint is scattered across your users' minds
• Balance: ego enough to believe in a billion-dollar outcome, humility enough to know you know nothing