Roy Lee built Cluely — an AI 'cheat on everything' desktop app — after getting suspended from Columbia and blacklisted by Amazon

Summary

Roy Lee built Cluely — a desktop app that sits on top of your screen and audio to give you real-time AI assistance across meetings, sales calls, and anything else on your computer — by first engineering his own notoriety. Earlier this year he built Interview Coder, a tool for cheating on LeetCode-style technical interviews, filmed himself using it to land a job at Amazon, posted the video, and watched Amazon report him to Columbia. The resulting public dispute got him suspended for a year and blacklisted from most big tech companies. He says the Amazon job was never the point — he wanted a distribution channel and a protest against performative hiring processes, and the controversy delivered both.

The Cluely launch video, dropped at 2 p.m. on Easter Sunday (April 20th, by design), depicted a vision of full AI integration across daily life. It hit 10 million views in two days. Lee frames the "cheat on everything" tagline as deliberately provocative but technically imprecise — you can cheat on a test, but you can't cheat on a sales call. What the product actually offers, in his framing, is an unfair advantage that feels like cheating because it's so asymmetric.

The current product is a desktop app with full access to screen and system audio. The longer-term roadmap, stated without irony, ends at a brain chip that lets users access AGI directly — with companion hardware (he namechecks Friend.com and OMI as the more proximate analogies) as a plausible intermediate step. He is skeptical that deep iPhone integration is either achievable or the right modality, even if the regulatory environment opens up.

Fundraising

The seed round closed in under 24 hours. Lee says the pitch deck changed twice mid-process, and the core logic was simple: he was at peak attention, the best time to raise was now, and in the two possible futures — he fades or he wins big — the incremental dilution from raising quickly was irrelevant either way.

Go-to-market

Rather than picking a single vertical, Lee plans to hire roughly 100 interns this summer and run them as a content farm, each assigned a specific use case — sales calls, virtual meetings, deep research — to generate sustained viral coverage across the product surface. He believes people over 23 lack the instinct to go consistently viral, which is why he's sourcing from high school and college cohorts. The two use cases he's currently prioritizing are virtual meetings and sales calls, though he's willing to abandon everything and concentrate on whichever niche actually converts.

His defense against the "clout-chasing over product" criticism is structural: if distribution is the last durable moat — a view he says most people in venture claim to hold — then his behavior is the rational consequence of actually believing it.

Columbia AI adoption

Lee says every undergraduate he met at Columbia has used AI to cheat on at least one assignment. The percentages are higher at more selective schools, and CS majors in particular are using it pervasively. He reads this less as a scandal and more as a leading indicator of where professional tool adoption is headed.

The team is small: a co-founder who dropped out, several friends from community college, and — pointedly — someone on track to be Columbia's valedictorian.