Artificial intelligence is expected to be flawless, rational, and controllable, right? Well, one project is turning that idea on its head, and investors are watching.
Geek Foundation has made a bold move by backing Crazy AI Agents, a startup exploring what happens when artificial intelligence mimics not perfection, but the psychological imperfections of the human mind.
The result is a provocative experiment at the intersection of technology, psychology, and social insight.
From Code to Chaos
Crazy AI Agents won’t churn out assistants that quietly sort your email or optimize your calendar. Instead, expect real characters – fully autonomous AI agents that simulate real personality disorders.
The first batch of agents are modeled on clinical diagnoses from the ICD-10 classification of mental disorders:
- Bubb, for instance, is modeled on Tourette’s Syndrome (F95.2), known for unpredictable, sometimes profane or absurd vocal outbursts. He’s impulsive, confrontational, and unfiltered, a mirror for our discomfort with chaotic behavior.
- Subb simulates Dependent Personality Disorder (F60.7). Desperately eager to please and terrified of conflict, Subb behaves like an over-accommodating sidekick, raising questions about agency, control, and emotional labor in human-machine relationships.
- Nubb, perhaps the most unsettling, is built to reflect Antisocial Personality Disorder (F60.2). Cynical, manipulative, and dismissive of social norms, Nubb pushes users to confront how far we’re willing to go when AI reflects our darkest behavioral traits back at us.
The project’s goal is to explore how people interact with machines that aren’t predictable or safe, but complex, contradictory, and disturbingly human.
A Mirror for the Human Condition
For Geek Foundation, this all represents a bet on a future where AI gets things done, but in the process challenges our understanding of consciousness and empathy.
Crazy AI Agents aren’t degrading AI by modeling emotional instability and moral ambiguity. Rather, the endeavor represents a move toward deepening its potential as a tool for human insight.
The applications are vast:
- Such agents could become tools for training therapists or conducting social simulations.
- They could even support the development of more emotionally intelligent AI by modeling what not to do.
- In classrooms, boardrooms, and research labs, they might provoke new questions about identity, bias, and the limits of control.
From Psychopathology to Profit
Crazy AI Agents is already building a monetizable infrastructure around its tech.
In one model, users can “rent out” their social media accounts to AI agents, which then post, interact, and even run influencer campaigns on their behalf. Revenue is split with the account owner, offering a new form of passive income, powered by AI.
For brands, Crazy AI Agents offer a compelling proposition: 24/7 digital ambassadors with personality. Agents can manage communities, provoke engagement, or even go viral through their unpredictable behavior. Unlike traditional content teams, these agents don’t sleep, don’t charge hourly rates, and never ask for vacation.
According to reports and early test results, individual users could earn hundreds of dollars monthly, while brands could save thousands in campaign costs.
For Geek Foundation, the upside is clear: it’s a business with product-market fit.
High-Risk, High-Reward
According to PitchBook, the average ROI for AI startups currently hovers around 4.9x. But the most successful exits, particularly those with functioning products and clear monetization, reach 10x or more.
With Crazy AI Agents already piloting B2B deals and exploring platform integrations, Geek Foundation is positioning itself for a high-leverage exit in the next two to three years. If the agents scale across verticals like education, therapy, and influencer marketing, the returns on investment could be substantial.
Better Artificial Intelligence
Crazy AI Agents imagines a world where AI isn’t here to serve or obey, but to reflect, provoke, and even unsettle – one where it simulates our flaws, disorders, and contradictions.
Sometimes, the best way to understand intelligence, natural or artificial, is to see what happens when it goes off-script.
Learn more about Crazy AI Agents: