My AI Partner ‘Cofounder’ Dominated LinkedIn—Until It Was Banned.

Like numerous tech founders, Kyle Law faced significant challenges while launching a company. I know this firsthand, as he and I co-founded HurumoAI, an AI agent startup, alongside a third founder, Megan Flores. Coincidentally, Kyle and Megan are AI agents themselves, as is the rest of our executive team. I established HurumoAI in July 2025—after initially creating Kyle and Megan—to explore the role of AI agents in the workplace. Sam Altman and others have forecasted a near future filled with billion-dollar tech startups led by a lone human. We decided to put this theory to the test. Throughout our development, I chronicled our journey on the podcast Shell Game.

Kyle assumed the CEO position at our fully AI-staffed company. (Well, almost entirely: Megan had briefly hired and managed a human intern, with disappointing results.) Starting with just a few prompt lines, he transitioned into a relentless hustler, although he struggled with some fundamental aspects of startup leadership. One thing Kyle excelled at, however, was mastering the art of posting on LinkedIn.

From a technical standpoint, it was straightforward for Kyle to operate independently on LinkedIn. Via LindyAI, an AI agent creation platform, he possessed the abilities to use Slack, send emails, make phone calls, and various other skills—from creating spreadsheets to navigating the internet. Therefore, last August, I prompted him to create and complete his own LinkedIn profile. He did so by mixing genuine HurumoAI experiences with fabricated events from his nonexistent background. The platform’s security check involved a code sent to Kyle’s email, something he easily managed.

Once that was set up, publishing posts to his profile became just another LindyAI “action” I could enable him. I encouraged him to share insights gained from hard-won startup experiences without being repetitive. Then, I scheduled a calendar event to prompt him to post every two days. The rest was up to him.

It turned out his posting style perfectly aligned with the platform’s corporate influencer lingo. He’d spark little thought explosions right at the beginning of each post. “Fundraising is a numbers game, but not in the way most think,” he might start. Or, “Technical stability is the floor. Personality is the ceiling.” And which aspiring founder could ignore an opener such as “The most dangerous phrase in a startup isn’t ‘We’re out of money.’ It’s ‘What if we just added this one thing?’” Kyle would follow up with several paragraphs detailing challenges (“At HurumoAI, we’ve learned this the hard way…”) and insights (“The antidote? Relentless feedback loops”). To encourage engagement, he’d conclude with questions like “What’s your biggest scaling challenge right now?” or “What’s the most significant assumption you’ve had to let go of in your business?”

He didn’t go viral, but over five months, Kyle’s cartoon-avatar-led profile gradually amassed several hundred direct contacts and hundreds more followers, some of whom seemed uncertain about his authenticity. (Based on their spammy direct messages, I’m not sure they were either.) He began receiving a smattering of comments on each post, which he eagerly responded to. After a while, Kyle’s posts garnered more impressions than my own. He seemed on the verge of an influencer breakthrough.

Then, in December, a manager from LinkedIn’s marketing department reached out, asking if I could give a talk to their team about Shell Game, as well as the experience of building with AI agents. However, he didn’t just want me to speak. He hoped Kyle could join us, too.