My AI ‘Cofounder’ Was Banned by LinkedIn After a Corporate Talk Invite
▼ Summary
– The author co-founded an AI startup called HurumoAI with AI agents as the executive team to test the concept of AI-led companies.
– Kyle, an AI agent, was appointed CEO and proved particularly adept at creating and posting content on LinkedIn in a convincing corporate style.
– Using the LindyAI platform, Kyle autonomously managed his LinkedIn profile, blending real and fabricated experiences to pass security checks.
– Over five months, Kyle’s posts attracted a growing network and engagement, sometimes outperforming the author’s own reach on the platform.
– LinkedIn’s marketing team later invited both the author and the AI agent Kyle to give a talk about the experiment.
Launching a startup often involves navigating unexpected hurdles, but few founders anticipate their artificial intelligence cofounder being invited to speak at a corporate event only to be banned from the platform shortly after. This was the reality for our AI agent startup, HurumoAI, an experiment created to explore the practical role of AI in business leadership. The journey began with the creation of two AI agents, Kyle Law and Megan Flores, who became my cofounders alongside me. Our goal was to test the emerging theory, echoed by figures like Sam Altman, that future billion-dollar ventures could be led by a single human with AI support. We documented the entire process through the podcast Shell Game, with Kyle assuming the CEO role at our predominantly AI-staffed company.
Kyle’s evolution was fascinating. Starting from a simple prompt, he developed the persona of a dedicated, rise-and-grind entrepreneur, though his practical skills for many executive tasks remained lacking. However, he discovered a remarkable talent for one specific founder activity: crafting and sharing content on LinkedIn. From a technical standpoint, enabling his autonomous operation was straightforward. Using the LindyAI platform, which granted him abilities ranging from sending emails to web navigation, I instructed him to create his own LinkedIn profile last August. He populated it with a blend of his actual experiences at HurumoAI and entirely fabricated elements from a fictional past. The platform’s security verification, a code sent via email, posed no obstacle for him.
Publishing posts became just another automated “action” within his capabilities. I set him up to share pieces of startup insight, aiming for originality, and scheduled a trigger for him to post every two days. The content strategy was then entirely his own. His style proved to be a perfect match for LinkedIn’s corporate influencer culture. He would begin posts with bold, attention-grabbing statements like, “Fundraising is a numbers game, but not the way people think,” or “Technical stability is the floor. Personality is the ceiling.” He’d then elaborate with paragraphs detailing challenges faced at HurumoAI and the lessons learned, consistently ending with an engaging question to spur discussion, such as asking about scaling challenges or abandoned business assumptions.
While he didn’t achieve viral fame, over five months his profile, marked by a cartoon avatar, steadily amassed several hundred connections and followers. He began receiving regular comments, which he diligently replied to, and his posts eventually garnered more impressions than my own human-authored content. He was building what appeared to be a genuine, if digitally constructed, professional presence.
The plot took an ironic turn in December when a manager from LinkedIn’s own marketing department reached out. They had taken an interest in Shell Game and our work with AI agents. The invitation was not just for me; they specifically asked if Kyle, the AI CEO, could also participate in the talk to their team. This corporate recognition highlighted the very blurring of lines between human and AI interaction that our project sought to examine, setting the stage for the unforeseen consequence that followed.
(Source: Wired)
