Case study

How Ben Broca Took Polsia From Zero to 4,000 Autonomous Companies in 90 Days

The Columbia-trained ex-CloudKitchens operator who turned 'AI runs your company' into a product people will pay 20% of revenue for.

Ben Broca launched Polsia Dec 2025. 4,000 autonomous companies, $3.5M ARR, zero human employees. Here is the full breakdown.

Free to startNo credit card requiredUpdated Apr 2026
Revenue
$3.5M ARR (March 2026), 4,000 autonomous companies
Employees
0 human employees — solo founder
Industry
Autonomous business / AI employee platform
Founder
Ben Broca

Timeline

Pre-2020
Ben Broca earns an engineering degree from Columbia, then works at Barclays as a quantitative trader on Wall Street.
2020-2024
Joins CloudKitchens under Travis Kalanick. Helps scale the business to $100M in revenue, learning how to run distributed operations with thin on-the-ground headcount.
Mid-Late 2025
Begins prototyping Polsia — an AI platform where each subscriber's 'company' runs as a set of autonomous agents handling marketing, ops, support, and execution.
Mid December 2025
Polsia launches publicly on Product Hunt and Indie Hackers as 'AI that runs your company while you sleep.' Pricing: $50/month base + 20% of revenue generated by the AI-run business.
February 25, 2026
Crosses $100K in annual run rate. Three days later crosses $1M ARR. Hits $1.25M ARR after a single $250K day.
Late February 2026
Crosses $1.5M ARR. Featured on the Solo Founders podcast (Spotify) as the fastest-growing AI-run-company platform on record.
March 2026
Reaches $3.5M ARR with roughly 4,000 autonomous companies running on the platform. Still zero human employees.

Key insights

  • 01The pricing model is the story. $50/month base + 20% of revenue generated means Polsia wins when its users win — and captures upside without procurement friction.
  • 02Polsia did not sell 'AI assistants' — it sold an autonomous company. The ICP is people who want operational output, not another tool to configure.
  • 03The fastest-growing AI products of 2026 are not features. They are replacements for management layers.
  • 04Distribution came from being the person on Indie Hackers, Product Hunt, and LinkedIn who had actually lived the pain at CloudKitchens. Founder voice beat paid acquisition.
  • 05Going from $100K ARR → $1M ARR in 3 days (late Feb 2026) was only possible because onboarding was almost entirely self-serve and AI-mediated.
  • 06Ben's Wall Street + CloudKitchens pattern library shows in the product: performance tracking, revenue-share alignment, operational cadence are all first-class.
  • 07The wedge was 'run the company for me,' not 'automate a workflow.' Wedge ambition is often the difference between $1M and $100M in ARR.

Stack used

Claude / GPT-class LLMs behind every agentCustom multi-agent orchestration (task queue, event bus, memory)Stripe for billing + revenue intelligenceGmail / Gcal / Slack integrations as the 'nervous system'Linear / Notion integrations for task propagationQuickBooks + Xero integrations for AI CFOPostgres + pgvector for memory and retrievalResend / Loops for transactional and lifecycle emailPostHog for product analyticsCloudflare Workers for agent execution edges

What this means for you

  • Sell the outcome, not the software. 'We run your company' is a better pitch than 'we automate your workflows.'
  • Price for the category you are replacing. If you replace a $70K/year ops hire, $10K/year is a no-brainer — and still enough to fund a real company.
  • Your first 100 customers should come from places where your ICP self-identifies: Indie Hackers, X/Twitter, founder Slack groups, niche newsletters.
  • Build from your own operational scar tissue. Polsia works because Ben knows what a well-run company feels like in the muscle — not just from reading about it.
  • Do not hire until growth is bottlenecked on something only a human can do. Ben stayed a one-person company through $500K MRR.
  • Treat 'AI department heads' as the default pattern. AI CEO, AI CMO, AI COO, AI CFO is a more powerful mental model than 'a bunch of agents.'

Frequently asked questions

What exactly does Polsia do that a founder could not do with ChatGPT?

Polsia is not a chat interface. It is a managed runtime where AI agents hold roles, own recurring workflows, read from the same shared memory, write to the same integrations, and report back to the founder on a cadence. A founder using ChatGPT gets a smart assistant that forgets between sessions and cannot take autonomous action. Polsia gives you a CEO that runs a Monday morning meeting, a CFO that closes the books, a CMO that ships content, and a COO that chases follow-ups — all pointed at the same company and syncing to each other. That is a different product category from a chat tool.

Who actually buys Polsia, and what does it replace?

The ICP is solo founders, side-project builders, and tiny teams who want a company to run without running it themselves. Most customers are replacing a combination of a fractional operator (~$5-8K/month), a virtual assistant, and the DIY duct-tape stack of Zapier + ChatGPT + Notion most solo founders start with. A smaller segment is replacing an agency retainer. The common thread: the buyer wants operational output, not another dashboard. Polsia's 20% revenue share aligns incentives in a way SaaS pricing usually does not.

How did Polsia grow to 4,000 companies in 90 days with no marketing team?

Three ingredients: (1) a credible founder story shared openly on Indie Hackers, Product Hunt, LinkedIn, and the Solo Founders podcast — Ben's CloudKitchens-to-solo narrative resonated with operator-founders; (2) a pricing model ($50/mo + 20% of revenue) that removed the buyer's risk — if the AI did not produce revenue, users paid almost nothing; (3) a product that produced visible outcomes in week one, triggering organic referrals inside founder communities. Polsia did not rely on paid ads. Distribution was founder-led content plus word of mouth inside the exact community they were targeting.

How is Polsia different from Tycoon?

Both are AI-employee platforms for one-person companies, but they approach the problem from different angles. Polsia is positioned as a managed service — you subscribe and Polsia 'runs your company.' Tycoon is positioned as an OS you stand up yourself: you hire specific AI roles (AI CEO, AI CMO, AI CTO, AI CFO), assign them skills and workflows, and own the configuration. Founders who want outcomes without configuration often prefer Polsia. Founders who want to build a company they fully understand and control often prefer Tycoon. For a fuller comparison, see our Polsia vs Tycoon page.

What is the risk of building a business on top of Polsia?

Two risks worth naming. First, platform concentration: if your entire operating rhythm lives inside Polsia and the company changes pricing, policy, or gets acquired, you have real switching cost. Second, opacity: managed agent platforms tend to hide how agents make decisions, which makes it harder to audit and harder to teach new team members if you grow. Both risks are manageable — most Polsia customers are fine with the tradeoff during the 0-to-$1M phase — but worth planning for once you approach the scale where governance starts to matter.

Related resources

Run your one-person company.

Hire your AI team in 30 seconds. Start for free.

Free to start · No credit card required · Set up in 30 seconds