Building Cluely

The episode opens with Erik Torenberg saying Elon is reaching out and Meta is offering a $1 billion acquisition offer. Roy Lee responds that six months earlier he was a random college kid in a dorm and now feels at the center of tech, while the more astonishing development is how correct his virality assumptions have been.

Cluely’s central bet is that distribution can discover the product faster than traditional product development can discover demand. The team wrote its first code 10 weeks before the episode, launched with a barely functioning product, and inserted sales calls into its videos as a test; Roy says that experiment produced “over a million dollars of enterprise revenue coming in.” With more than 1 billion Cluely views, aggregate usage now points the team toward its stickiest uses and product direction.

The discussion frames Twitter as roughly two years behind Instagram and other platforms in understanding short-form virality, while Roy’s broader claim is that people on X and LinkedIn are behind. Those platforms reward accessible, controversial content, while tech creators optimize for intellectual status and produce material “maybe like 200 people in the world can actually understand.” His supposedly extreme videos are less controversial on Instagram or TikTok, where creators compete with content insinuating public felonies.

Cluely has reorganized marketing around demonstrated algorithmic mastery rather than conventional credentials. Bryan Kim puts Roy in roughly the top 0.1% in the world at distribution. Every full-time employee has over 100,000 followers on some social platform, while more than 60 contractors are paid per video to produce clips—including batches of five 10-second videos—that can generate millions of views. Roy’s provocation: if a head of marketing lacks 100,000 followers, “you need to replace them.”

Bryan backed Cluely after seeing it turn attention into dollars, but his larger thesis required moving beyond his old consumer-investing framework while retaining some belief in its core. He previously favored slowly crafted products with retention and network effects; rapidly changing AI models made that insufficient because an incumbent’s next release can erase a feature. In this phase, founders must enjoy “building the plane as it’s falling down the cliff,” making momentum across product and distribution the current moat.

The product thesis is that AI should inhabit a translucent overlay that sees the screen and hears audio, not remain trapped in a separate window. The UX emerged from 20–30 Interview Coder iterations, when Roy and co-founder Neil needed to view generated code and their own work simultaneously. Roy frames Cluely as “pre-launch”: saturate the market with the category now, then release the fuller product to an audience already primed to associate invisible AI with Cluely.

Roy concedes the overlay is technically simple and likely to be copied, so defensibility currently rests on winning a land grab. He argues Cluely might distribute better than OpenAI and says there is enough of a case to “probably bet on us” at roughly a “30,000× discount,” while controversy makes its marketing anti-fragile: attacks create supporters, opponents, and more attention. His guardrails are narrower than “triple down on everything”—“never punch down,” remain authentic, and let genuine respect remain visible.

🔗 Original source & video: Building Cluely: The Viral AI Startup that raised $15M in 10 Weeks w/ Roy Lee