How Roy Lee Turned an Interview Cheat Hack into a Legit Business Tool
By JL Zhang | 07 Aug, 2026

Lee and co-founders Neel Shanmugam and Alex Chen want Cluely's desktop overlay to make your business meetings more successful.

Cluely began with a proposition designed to make employers recoil: What if you could secretly use artificial intelligence to cheat your way through a job interview?

Its creators didn’t soften the idea. They demonstrated it during real technical interviews and later widened the provocation into a promise to help users “cheat on everything.”

That made Chungin “Roy” Lee one of Silicon Valley’s most talked-about young founders. It also left Cluely with a problem. A tool associated with deception might attract students and millions of social-media views, but purchasing departments don’t allocate budgets for cheating.

Lee and cofounders Neel Shanmugam and Alex Chen have spent the past year turning the same technology into something companies can legitimately buy: a desktop assistant that listens to meetings, understands what’s on the screen and supplies notes, answers and useful information while the conversation is still happening.

The conversion isn’t complete. Cluely still advertises an overlay that can remain invisible during screen sharing. Yet its shift toward sales, support, recruiting and consulting gives the technology a real business purpose beyond sneaking answers into interviews.

Roy Lee Arrives Looking For A Cofounder

Lee, a Korean American raised around Atlanta, attended Diablo Valley College in California before transferring to Columbia University to study computer science.

He has said one of his main goals upon arriving was finding a cofounder. He met Shanmugam near the beginning of the fall 2024 semester, and they began experimenting with side projects almost immediately.

Lee brought an instinct for product simplicity, provocation and online distribution. His thesis was that AI was making software easier to build, leaving attention as the scarcer resource. A tiny startup couldn’t outspend Microsoft or Google, but it might out-entertain them.

That belief would shape Cluely almost as much as its code.

Neel Shanmugam Brought Academic Firepower

Shanmugam grew up in Richardson, Texas, attended the Texas Academy of Mathematics and Science and entered Columbia as an Egleston Scholar.

Before college, he had worked on environmentally safer solvents, molecular simulations of cancer-related proteins and computational design of antiviral therapeutics. Columbia credited him with coauthoring nine peer-reviewed papers.

He initially imagined a future in biotechnology. At Columbia, however, he became interested in startups and shared Lee’s skepticism toward technical recruiting.

Software engineers use search engines, documentation and AI assistants while working. Job candidates are nevertheless expected to solve memorized algorithm puzzles without those tools. Lee and Shanmugam saw that mismatch as absurd—and commercially exploitable.

Their answer was Interview Coder.

The Hack That Started Everything

Interview Coder was a translucent desktop overlay designed to remain hidden from interviewers and screen-sharing software. It could view a coding problem, listen to the interview and generate suggested solutions in real time.

Lee used it in actual interview processes and posted recordings. The videos spread rapidly, helped by claims that the tool had assisted him during interviews with major technology companies.

Priced around $60 a month, Interview Coder was reportedly nearing a $2 million annual revenue pace by March 2025. It had uncovered genuine demand, even if that demand depended on conduct most employers considered dishonest.

The publicity attracted Columbia’s attention. The disciplinary story was more complicated than the claim that Lee was expelled merely for building the app. His eventual one-year suspension cited the publication of materials from a confidential disciplinary proceeding.

Shanmugam left before his own process reached the same point. Lee dropped out rather than wait through the suspension.

Instead of killing the business, the controversy gave it rocket fuel.

Alex Chen Turned A Hack Into A Platform

Alex Chen joined as the technical cofounder who could help widen Interview Coder into a general desktop assistant.

A Columbia computer science student, Chen had worked at Teamflow, a virtual-office startup, and at AI-agent company Lindy. His experience included collaboration software, language-model evaluation and practical AI assistants.

Cluely needed that breadth. A coding-answer generator only had to recognize technical questions. A business assistant had to interpret many applications, combine audio with screen context, retrieve company information and respond quickly enough for live conversation.

Public accounts offer fewer details about precisely when Chen formally joined. His role nevertheless created a logical division of labor: Lee became chief executive and distribution strategist; Shanmugam became operating cofounder; and Chen led the broader technical platform.

Cluely Launches By Courting Outrage

The trio launched Cluely in April 2025 with $5.3 million in seed financing led by Abstract Ventures and Susa Ventures.

Its launch video showed Lee on a date using the assistant to misrepresent his age, career and knowledge of art. The product arrived under the slogan “cheat on everything.”

The campaign was outrageous, instantly understandable and perfect for social media. Roughly 70,000 people reportedly signed up during the first week.

Cluely treated attention almost like a manufacturing process, recruiting creators, editors and growth interns to produce streams of short videos. Lee argued that employees should either improve the product or make it go viral.

The strategy attracted Andreessen Horowitz, which led a $15 million Series A only two months later. Total disclosed financing reached $20.3 million, with investors estimating a post-money valuation around $120 million.

That proved Cluely could attract attention and capital. It didn’t prove users would keep paying after the novelty faded.

What Cluely Does Now

Cluely now presents itself primarily as an AI meeting assistant.

The desktop application listens to a call and examines relevant information on the user’s screen. It can transcribe the conversation, produce notes, detect questions and generate suggested answers inside a movable overlay.

Unlike notetaking bots that visibly join a Zoom or Teams call, Cluely operates from the user’s computer. There’s no extra bot on the guest list.

During a sales call, it might surface an approved response to a pricing objection. During a customer-support session, it might retrieve a troubleshooting step. Afterward, it can summarize the conversation, identify action items and draft a follow-up email.

Enterprise customers can load documents, scripts and operating procedures into a knowledge base. Cluely retrieves relevant material as the call unfolds and can connect with customer-relationship systems.

That’s the legitimate business case: reducing the burden of remembering every detail while giving employees immediate access to what their organization already knows.

Most meeting assistants concentrate on what happens after a call. Cluely tries to help during it.

A salesperson can’t wait until afterward to discover the best response to an objection. A recruiter may need a revealing follow-up immediately. A consultant may need a number from a long report without stopping the conversation to hunt for it.

The overlay keeps the user inside the natural flow of the meeting instead of forcing constant switching among a chatbot, browser tabs and company documents.

The same invisibility also creates distrust. Other participants may not know their words are being processed or recorded. Consent laws vary, and employers may prohibit concealed assistance even outside interviews.

A legitimate business deployment therefore requires transparency, clear data-retention policies and agreement about when AI assistance is allowed.

The Product Had To Grow Up

Early independent tests found Cluely slow, distracting and prone to generic answers. Suggestions sometimes arrived too late to use naturally. Lee acknowledged that the first release was extremely raw and may have launched too early.

That was partly intentional. His strategy was to release something barely functional, attract a huge audience and let users reveal the most valuable applications.

That approach can work for consumer software, but it’s risky for a tool meant to assist during high-stakes conversations. A response arriving 10 seconds late may be useless. A confident hallucination about pricing, policy or a technical specification could be worse than offering no answer at all.

By late 2025, the company had narrowed its pitch. References to cheating on dates, exams and interviews receded. Meeting notes, business answers and follow-up actions moved to the center.

Cluely now offers a free tier, a Pro subscription around $20 a month and a $149.99 monthly tier advertising invisibility during screen sharing. Enterprise plans add knowledge bases, integrations, controls, coaching and security features.

The repositioning places Cluely in a crowded market alongside Otter, Fireflies, Granola, Gong and many specialized AI sales assistants. It reduces the shock value but makes the product easier for a manager to expense.

“Cheating on everything” may be a brilliant social-media hook. It’s a terrible description to place on a corporate purchase order.

The Numbers Behind The Noise

Cluely’s business activity is substantial for a company barely more than a year old, but its figures need scrutiny.

It raised $20.3 million in two disclosed rounds. The reported 70,000 registrations during its first week showed extraordinary curiosity, though signups aren’t the same as paying or retained users.

Lee later claimed Cluely had reached $7 million in annual recurring revenue. In March 2026, he admitted that number was false.

The figures he subsequently released showed about $2.7 million in consumer ARR and $2.5 million in enterprise ARR at the relevant time, or roughly $5.2 million combined. That equals about $433,000 in recurring monthly revenue.

Company documentation says Cluely serves more than 1,000 enterprise seats. The company hasn’t released enough information to calculate how many paying consumers it retains because its customer mix includes free users, annual discounts, standard subscriptions, the expensive invisibility tier and negotiated enterprise contracts.

Using the corrected revenue figure, the reported $120 million valuation was approximately 23 times ARR. Cluely had also raised about $3.90 for every dollar of annualized revenue.

Those aren’t bad ratios for a young AI startup, but they don’t reveal retention, churn, customer-acquisition costs or gross margins. Viral signups become a durable business only when customers keep using the product after the spectacle ends.

The nearly even split between consumer and enterprise revenue was encouraging. It suggested Cluely wasn’t relying entirely on curious individuals trying the controversial interview tool.

It also underscored the importance of its business pivot. Enterprise contracts generally offer larger accounts and lower churn than individual subscriptions, but corporate buyers demand stronger security, administration and reliability.

Trust Is Now The Real Product

Cluely’s greatest early strength was its willingness to offend. Its greatest long-term weakness may be the distrust that created.

Companies buying software that hears conversations and reads screens need confidence in security, consent and accuracy. Lee’s admission that he inflated revenue is particularly damaging when the company is asking customers to trust it with sensitive calls.

The product also relies on outside AI models. When those models hallucinate, Cluely can surface confident but wrong advice precisely when a user is under pressure to respond.

The software may be most useful when it retrieves approved information from a company’s own knowledge base rather than improvising general answers. A verified price, policy or contract provision is more valuable during a meeting than an eloquent guess.

Cluely’s future therefore depends less on staying undetected than on becoming fast, dependable and accountable.

A Legitimate Company Born From Cheating

Interview Coder succeeded because it attacked a narrow pain point with shameless clarity. Cluely must now succeed for almost the opposite reason: it has to become respectable enough for organizations to deploy openly.

Lee, Shanmugam and Chen have shown they can build quickly, capture attention and convert some of it into paying customers. What remains uncertain is whether they can create a product people continue using after the controversy stops being entertaining.

The irony is that Cluely’s best future may have little to do with cheating.

Employees enter meetings surrounded by more information than anyone can remember. Sales representatives can’t memorize every product detail. Consultants can’t hold every page of a client’s documents in their heads. Managers routinely forget promises made during fast-moving discussions.

An assistant that finds the right fact, preserves commitments and suggests a useful question could make those meetings genuinely better.

To complete the transformation, Cluely will eventually have to stop treating invisibility as its defining achievement. The real prize isn’t helping users pretend they know more than they do.  It’s helping them use what their organizations already know at the exact moment it matters.