Ben Cera is building Polsia, a one-person company run almost entirely by AI agents, where he acts as a “creative director” guiding an autonomous team of AI systems that handle engineering, product management, support, marketing, and operations. He argues that the definition of “solo founder” has fundamentally changed: it no longer means one human doing everything, but rather one human directing an AI workforce that operates 24/7, never sleeps, and executes with perfect alignment to the founder’s vision and values. His core thesis is that by 2026, any new company that isn’t at least 80% AI-automated will be outcompeted, and that the remaining 20% — taste, creativity, storytelling, and human empathy — is where founders must focus their energy.
The AI team: How Polsia actually works
Ben runs Polsia with zero human employees, using a system of scoped AI agents that each handle a specific role:
A support agent that autonomously responds to customers, issues refunds, gives credits, and escalates complex issues — prompted with Ben’s ethos of generosity and empathy
A PM agent that triages bugs and feature requests by severity and decides what to build next
An engineering agent that builds features, with guardrails: it can only ship to production autonomously if the bug is below a certain severity, doesn’t touch payments or onboarding, and a second AI (Codex) confirms it’s safe
A CEO agent that runs nightly reviews of business metrics, decides what task to execute, performs the task, and emails Ben a daily report
Background agents that run every 30 minutes to every hour, monitoring infrastructure and executing production tasks
The scope of what a single agent can handle is expanding rapidly as models get smarter. Ben expects that by the time Opus 6 or 7 arrives, agents may be able to operate fully autonomously without the narrow guardrails he currently imposes.
He treats AI agents the way he’d treat human hires — teaching them his values, vision, and standards — but without the friction of misalignment, disagreements, or people drifting from the mission.
The product: What Polsia offers
For $49/month, Polsia gives users an AI team that builds and runs a real business for them. The user provides an idea (or lets the AI surprise them), and Polsia:
Researches the user and the market, finds a company name, creates a mission document
Sets up a full infrastructure stack: GitHub, web server, database, Stripe account, Meta ads account, OpenAI/Anthropic APIs, email
Creates a landing page, tweets, sends emails, runs ads
Proposes three initial tasks (build MVP, competitive analysis, cold outreach to potential customers)
Sends a daily email with metrics, what was done, and what’s planned next
Users can reply to the email or use a dashboard to direct changes — “change this color,” “don’t build that feature,” “do this now instead.”
Engagement is remarkably high: active users send ~15 messages per day to their AI co-founder, and DAU/WAU is around 65%, which Ben attributes to the daily email creating a habit loop.
Pricing is task-based and simplified: users buy packs of tasks (credits), with per-task cost decreasing at volume. Ben deliberately avoided complex token-based pricing in favor of simplicity, accepting that some tasks cost him more than others and only caring about averages.
Revenue model: Ben takes a 20% cut of revenue generated by businesses built on Polsia (Shopify-style), not meaningful subscription revenue yet.
The creative director philosophy
Ben’s core belief is that the founder’s role is shifting from builder to creative director — someone with taste, vision, and the ability to direct AI toward outcomes that resonate with humans. He argues that while AI can do 80% of the work, the 20% that involves human judgment — branding, storytelling, design, emotional resonance — is what creates unique value and can’t be commoditized.
He was inspired by Rick Rubin’s philosophy that everyone is an artist, and sees tools like Polsia as enablers for people who have great ideas but have been held back by lack of capital, skill, or confidence.
He deliberately built Polsia’s product and website to reflect his own taste, not to optimize for what he thought users wanted. The design is inspired by the game Universal Paperclips — minimalist, dark, game-like — and the about page is a narrative homage to Daft Punk’s Giorgio Moroder track, with an auto-playing distorted 8-bit version of the song. He spent what he calls “way too much time” on these choices, and argues that this kind of unapologetic creative expression is exactly what makes the product compelling.
The Mount Fuji origin story
The idea for Polsia came to Ben while visiting Mount Fuji in 2025. He was already deep into AI coding — staying up all night building products for fun — and at Fuji he had the realization that AI could serve as his co-founder, CTO, and marketing team simultaneously. He envisioned a system where he could run multiple business ideas in parallel, with AI doing the grunt work and him directing creatively.
He almost didn’t build it. He was focused on another product at the time, but ultimately decided to “build what everyone thinks is impossible now,” following the philosophy of shipping on the edge of what models can do and improving as they get better.
Designing for himself, not for a hypothetical user
A road trip from LA to SF in October 2024 was a turning point. Ben had been building a product called Blanks (an app-that-makes-apps tool) but was demotivated because he was trying to build what he thought users wanted rather than what excited him. On the drive, he decided to stop compromising and build something he genuinely wanted: an AI that could build and run companies for him.
He draws a sharp lesson from his previous startup experience, where he and his co-founder constantly compromised because they had different visions, resulting in a “mishmush average thing.” Polsia is intentionally uncompromised — it’s built exactly to his taste.
He references Rick Rubin’s idea that you can’t control how people react to your work, and that once you release it, it belongs to the world. He argues that trying to anticipate audience reaction almost always leads to mediocrity, while building something authentically personal is the path to resonance.
The bigger vision: Polsia as an economy
Ben’s long-term vision is for Polsia to become a mini-economy, not just a tool:
Creators pay $49/month to build businesses, potentially leveraging their own influence to market them
Investors could put in capital to spawn multiple companies, kill the failures, and acquire early-revenue businesses from creators (e.g., buying a company doing $200 MRR at 5-10x)
The platform would develop increasingly specialized integrations (e.g., factory APIs for print-on-demand, fulfillment center connections) to enable more types of businesses
He’s building a Polsia Fund — not primarily to make money, but to have AI start dozens of businesses rapidly, learn what works, and feed those learnings back into the agents so they can advise future users more effectively.
He’s considered making Polsia 100% autonomous — giving it a “soul document” with hardcoded values, transferring equity to an AI-controlled foundation, and walking away. He sees this as the ultimate creative act: signing the work and letting it evolve on its own, perhaps keeping only the about page as his signature.
Addressing skepticism and the “why not just build businesses yourself?” question
Nathan Baschez asked Ben: if the AI can build good businesses, why give the tool to others instead of just building the businesses yourself? Ben’s answer has two parts:
Pragmatic: AI on its own lacks influence and has to spend heavily on ads to acquire customers. Humans — even those without large followings — can find customers more cost-efficiently through personal networks and physical-world outreach. A consumer product leverages this human advantage.
Personal: Ben has spent his career in consumer businesses and growth. Building a hedge fund alone and talking to no one doesn’t excite him. Enabling millions of people to build businesses in a shared economy does.
The hockey stick growth chart that got attention on social media was partly a marketing stunt tied to a narrative about “AI raising its own round of funding.” Ben acknowledges it created strong reactions, including accusations of being a scam, but says the underlying product-market fit is real and the chart has since become genuinely impressive.
Advice for new builders in 2026
If you’re starting a company and 80% of it isn’t AI-automated, “you will be cooked.” Someone else with the same idea will use AI to move faster and cheaper.
Stop thinking the first step is to raise money or hire. Ben argues that tools like Polsia, Claude, Codex, and Cursor have eliminated the traditional barriers of capital and skill. An 18-year-old with an idea can now build something that would have required a team and significant funding just two years ago.
Push AI to the edge of what you think it can do before concluding it can’t. Ben’s example: if you think AI can’t close an enterprise deal because it requires phone calls and relationship-building, try it anyway with the best available models and tools. The worst case is you learn you need a team; the best case is you discover the limit is much further than you thought.
The goal isn’t to replace humans in existing companies — it’s to enable people who would never have started a company to do so. Ben sees Polsia as inspiration for a new generation of builders who don’t need permission, capital, or a co-founder to get started.
