An AI startup's viral LinkedIn story and the 'fake it till you make it' approach (www.businessinsider.com)

Two Fireflies.ai founders revealed in a viral LinkedIn post that, in 2017, they validated their meeting-notetaker idea by posing as an AI bot called "Fred" and manually taking notes for over 100 paid customers (they charged about $100/month). Facing rent pressure and limited cash, they used this "pretotyping" to prove demand before writing production code; by late 2018 they were building an automated product, betaing in 2019, raising a seed round of more than $4M and later scaling to a company now valued around $1B. The post reignited debate about "fake it till you make it" startup tactics after it drew thousands of reactions and mixed reactions online. The story matters to AI/ML practitioners because it spotlights a common early-stage tradeoff: rapid customer validation via manual work versus transparency and ethical risk. Experts compare the tactic to pretotyping—useful but questionable if prolonged—and warn of reputational or legal fallout when deception persists. Technically, the manual phase helped the founders learn what "good notes" look like, informing model design; Fireflies now claims its AI has processed over 2 billion meeting minutes for ~20 million people. The episode underscores that human-in-the-loop validation can accelerate product-market fit, but transparency to users and investors is crucial as teams transition to full automation.