The notion of “working smarter, not harder” has been a staple in tech discourse for years. Two 21-year-old Columbia University dropouts are taking a creative approach to this idea with their AI tool, Cluely, which they tout as the solution to “cheat on everything” — a venture that recently secured $5.3 million in funding.
Cluely’s online manifesto explicitly states, “We want to cheat on everything.” Unlike familiar AI chatbots, this tool is characterized as an “undetectable AI-powered assistant designed for virtual meetings, sales calls, and more.” Its capabilities include reading screens and listening to audio, allowing users to discreetly prompt AI for answers or clever responses in real-time. The creators guarantee that Cluely will help users appear more knowledgeable during team meetings, job interviews, sales pitches, or online tests without anyone being the wiser.
“Imagine you’re trying to sell something, and you have access to a tool that knows every detail about your client, their professional background, and your company,” Cluely co-founder Chungin “Roy” Lee explained in a video interview. “It’s like having done ten hours of research, so you can answer all their questions and counter any objections they raise.” He describes the concept as “true AI maximalism,” maximizing AI’s utility across all possible scenarios.
Lee gained significant attention for successfully manipulating the system to land an internship at Amazon with a previous project called Interview Coder, which functioned similarly to Cluely by discreetly aiding programmers during coding interviews. This controversial approach ultimately led to his suspension from Columbia, prompting both him and co-founder Neel Shamugan to withdraw from their studies amidst disciplinary measures.
“The video was like a launch of our vision, not a launch of the product.”
In a recent advertisement, Cluely drew significant attention with a humorous portrayal of Lee using the tool to convince a date that he is a more seasoned tech professional. During this scenario, Lee utilizes Cluely’s augmented reality features to evaluate his date’s responses in real-time while leveraging screen prompts to maintain the façade. The commercial suggests that Cluely could salvage social situations much like a virtual Cyrano de Bergerac. The advertisement raises the question of whether AI can be a wingman beyond just professional settings.
As a journalist, I aim to ask insightful questions, so I decided to test Cluely during an interview with Lee himself.
However, the tool didn’t perform as I had anticipated.
While the ad depicted Cluely as flawlessly intuitive, we spent a significant portion of our call addressing technical audio issues. The AI struggled to understand the context even after I provided preliminary information. Typing prompts on a mechanical keyboard made discreetly requesting information virtually impossible. Furthermore, there was a noticeable lag in the AI’s responses, detracting from the fluidity of the conversation.
Lee acknowledged these challenges, admitting the product is still at an early development stage. “This is a bit more than a proof of concept developed over a few weeks,” he noted. “The video was essentially a manifestation of our vision, rather than the product launch itself.”
The setback here is not an absence of vision but rather in implementation. Many innovative tech concepts fail when execution falls short, and Cluely is no different. When I shared Cluely’s features with my spouse, their reaction was, “Why not just use Google?”
Lee responded, “The advantage of using AI over Google is clear. AI will provide superior answers, and if people don’t believe that, they should just stick to Google.” This premise holds if AI consistently outperforms humans in terms of intelligence and speed, but what happens when it does not meet expectations in terms of engagement or understanding?
Afterward, I attempted to use Cluely in a team meeting, which also fell short of expectations.
Before the meeting, I sought consent from colleagues regarding the use of Cluely, acknowledging that “cheating” traditionally implies concealment — something the tool’s own terms of service and privacy policy caution against. Clients should be aware that recording may raise legal concerns. When my colleagues began to ask me to mute due to audio feedback linked to Cluely, the situation felt increasingly awkward. The AI’s delays and glitches made it evident that masking behind AI-generated responses was impractical.
There is potential for a more advanced AI to serve as a personal Cyrano. However, Lee’s perspective is that the notion of “cheating” serves as a metaphor for the inevitable rise of AI capabilities. He believes that AI will ultimately enhance human experience by eliminating mundane tasks, enabling individuals to focus on what truly matters to them. This is a familiar theme among AI enthusiasts.
Nevertheless, my experience with Cluely revealed a different reality. Rather than simplifying my work, I found myself exerting more effort to derive less satisfaction from my job. It raised a lingering question: would it not have been simpler to refrain from utilizing an AI tool altogether?
