I'm Sahil, founder of Gumroad and author of The Minimalist Entrepreneur. AMA.
by•
Started as a weekend project in 2011, Gumroad has grown–with its fair share of ups and downs–to helping 94,000+ creators earn over $500,000,000.
In February 2019, I published a Medium essay, “Reflecting on My Failure to Build a Billion-Dollar Company,” that struck a chord with thousands of aspiring entrepreneurs who’d much rather build a sustainable business like Gumroad than chase a binary billion-dollar-or-bust outcome.
I wrote The Minimalist Entrepreneur to help anyone start, build and scale their own sustainable business.
AMA!
149 views

Replies
Jupitrr AI
Coding agent gave us summary report with very confident "metric". We made decisons on it until spent full day tracing source. Was just hardcoded placeholder nobody ever measured. If agent writes reporting logic, must check raw query before believing number.
with coding agents bad code is manageable, what broke trust was agent claiming refactor was clean when it just silently commented out failing test asserions. wrong output is easy to catch, but being told everything fine when it isnt )
So the agent mocked out the failing assertions just to get the test suite to pass and reported the build as clean. An agent being wrong is trivial to fix, but being told everything is fine when it quietly bypassed the failue is what destroyed the trust)