Building a $1.5M ARR company alone using AI agents as your team

Executive overview

Most founders assume building a company requires hiring a team. Ben Cera hit $1.5M ARR with zero human employees by treating AI agents as full team members — not assistants.

The shift is architectural: instead of using AI to help you work, you design AI agents with defined scopes, tools, and decision authority to run operations, engineering, support, and marketing autonomously. The human role becomes creative director — setting taste, direction, and values — while AI executes.

The 20% that remains human is taste, creativity, and the ability to delight people who still hold the wallet.

What "solo plus AI" actually means

• Agents run every 30–60 minutes in production, checking infrastructure and executing tasks
• A support agent handles refunds, credits, and escalations autonomously using the best available model
• A PM agent triages bug reports and feature requests; an engineering agent builds and ships within defined constraints
• Scope limits replace human oversight: agents ship to production if severity is low, payments aren't touched, and a second model confirms safety
• Teaching an agent your ethos and values mirrors onboarding a human hire

How Polsia works

• User signs up and chooses: start a new business or get help with an existing one
• New business fork: choose a specific idea or "surprise me" (AI researches the user's background and proposes one)
• In 3–5 minutes: company name, email address, landing page, mission document, market research, competitor analysis
• On free trial activation: GitHub repo, web server, database, Stripe account, Meta ads account, and key API integrations are provisioned automatically
• Three pre-made tasks are created: build MVP, competitive analysis, cold outreach to potential customers
• Every night a CEO agent reviews logs, traffic, emails, and ad performance; executes tasks; sends a morning email summary with tomorrow's plan
• Users respond by email or via dashboard; active users average 15 messages per day, DAU/WAU is 65%
• Pricing: $49/month subscription plus task credits; Polsia takes 20% of revenue generated on the platform

The creative director model

• 80% of operations, engineering, and marketing can be automated; the remaining 20% is taste and direction
• "Surprise me" companies often outperform user-directed ones — less emotional attachment leads to less micromanagement
• Design choices at Polsia came from personal obsession (a minimalist clicker game, a Daft Punk song on the about page) not market research
• Building for yourself, not a hypothetical user, produces more distinctive outcomes
• Once you stop optimising for what others expect, the product becomes genuinely differentiated

The business model and long-term vision

• Current: consumer SaaS ($49/month) with revenue share, not pure subscription
• Short-term: Polsia fund — run AI-built companies internally to learn what works, feed learnings back into agent models
• Long-term vision: a mini-economy where users build businesses, investors acquire them, and successful companies provide real signal
• Future integrations could enable physical-world businesses (e.g., dropshipping, print-on-demand) via factory APIs
• Path to near-full autonomy: agents can already ship code and handle support; the missing pieces are hiring contractors and Instagram/social integrations

What to take from this as a builder

• If AI can't do something you think it can't, try it first — the limit is further than you assume
• An 18-year-old with a good idea and Polsia (or equivalent) has fewer excuses than any prior generation
• The cost of starting a company — capital, skill, time — is approaching zero; what remains scarce is knowing what you want
• Push agents to the edge; models will improve and make previously impossible things work
• The new definition of solo founder is "solo plus AI" — one human, many non-human teammates