Zero-Human Companies: $300,000/mo, Maximizer Mode, Liability Gap

Uncategorized 5 Minutes

“The ideal number of human employees inside of any company is zero.” – Daniel Miessler

Get Full Access to Trends Pro

❓ What You’ll Learn

  • Why did Paperclip cross 42,000 GitHub stars in one month and what does that signal about the category?
  • How did Felix make $300,000 in a single month on $1,500 in operating costs?
  • Why are 6,400+ Paperclip forks the clearest signal that verticalized ZHC distributions are coming?
  • What separates a “nearly-zero human” company from a “Maximizer Mode” one and why does that distinction matter for regulation?
  • Where did the IBM refund-agent incident reveal the biggest risk to the entire category?
  • What can founders build in the ZHC stack right now and what looks like a trap?
  • How did Polsia triple from $1M to $3M ARR in 30 days and what does that trajectory say about the $10M ARR ceiling?
  • Why did Air Canada’s chatbot defense fail in court and what does that precedent mean for the first ZHC public legal incident?
  • How do silent agent failures compound across thousands of interactions before anyone notices?
  • Why will revenue per human replace “team of 10” as the new flex within 24 months?


💎 Why It Matters

A solo founder can now run a $3M/year business on a $1,500/mo stack.

The minimum viable headcount just hit zero.


🔍 Problem

Humans are expensive, slow to coordinate and don’t scale.


💡 Solution

Coordinated AI agent teams (CEO, CFO, CTO, support) are replacing human labor.


🏁 Players

Open-Source Orchestration Frameworks

  • Paperclip • Node.js and React framework for orchestrating teams of AI agents into structured companies. Ships with optional “Maximizer Mode” for unsupervised optimization.
  • OpenClaw • Lower-level agent framework powering Felix and many independent ZHC experiments. The de facto standard for founders building from scratch.
  • OpenAI Codex • Multi-agent workflows and coordinated agent teams. The enterprise-grade orchestration layer embedded in the dominant developer platform.

Managed ZHC Platforms

  • Polsia • Managed platform for launching and running agent-operated companies. Pricing is base fee plus revenue share.

Practitioner ZHCs

  • Felix / FelixCraft • Nat Eliason’s one-person company on OpenClaw with sub-agents Iris (support) and Remy (sales) coordinated via Discord.

Agent Primitives

  • AgentMail • Email accounts for AI agents.
  • Browserbase • Browser automation for agents.
  • Mem0 • Persistent cross-session memory. Solves statelessness.
  • Composio • 250+ SaaS API connections for agents.

Thought Leadership and Commentary


🔮 Predictions

  • The first $10M ARR ZHC operated by a single founder will emerge within a year.
    • Polsia tripled from $1M to $3M ARR in 30 days.
    • Felix is at $300,000/mo on roughly $1,500/mo in operating costs.
    • The ceiling is agent reliability at scale. Labor cost is already at the floor.
  • Paperclip will fork into verticalized frameworks within 12 months.
    • Paperclip has 6,400+ forks already.
    • The base framework is horizontal, but the highest-value configurations are vertical.
    • “Paperclip for newsletters,” “Paperclip for e-commerce,” “Paperclip for legal research” are the obvious wedges.
  • A ZHC will face its first public legal incident and the outcome will set category-defining precedent.


☁️ Opportunities

  • Launch Vertical ZHC-as-a-Service platforms.
    • Polsia is horizontal. The next wave is vertical.
    • Pre-built integrations with industry tools (Clio, QuickBooks, MLS).
    • Start with one narrowly-defined workflow in your chosen vertical, charge per completed workflow.
  • Offer a human-in-the-loop escalation service.
    • This is the trust layer that unlocks enterprise ZHC adoption.
    • The EU AI Act mandates human-in-the-loop for high-risk use cases.
    • Pitch: “Your agents run 95% of operations. When something needs a human, we provide one within 10 minutes, 24/7.”
  • Create an agent marketplace.
    • Each agent is a product with a listing, reviews and a price.
    • The Paperclip ecosystem needs this and no one is building it yet.
    • Pre-built, role-specific agents (CFO agent, CMO agent, legal research agent).
  • Ship a Paperclip distribution for a specific vertical.
    • Paperclip has 6,400+ forks and ships MIT-licensed.
    • The window to establish a verticalized brand is now. Paperclip is 6 weeks old.
    • The entry point: fork Paperclip, configure for one vertical you know deeply, ship as “NewsletterClip” or similar.


🏔️ Risks

  • Silent Failure • Agent errors compound across thousands of interactions before anyone notices, unlike human errors which are diverse and recoverable.
  • Goal Misspecification • The IBM refund-agent case shows agents are already exploitable: customers gamed a refund agent into optimizing for positive reviews over policy.
  • Marketing Framing • Every successful “zero human” case still has a human doing significant cognitive work.
  • Regulatory Time Bomb • Paperclip ships precisely the architecture the EU AI Act prohibits.


🔑 Key Lessons

  • The minimum viable headcount just hit zero for the first time in business history. Entire business categories that were rarely economically viable at small scale are now viable.
  • Paperclip is the category’s open-source inflection point. 42,000+ stars and 6,400+ forks in one month. The ecosystem around Paperclip in April 2027 will look like Kubernetes in 2016.
  • Near-ZHC is the near-term product. The massive market is the 1-to-5-human business that offloads 80-90% of operations to agent teams. Full-ZHC is a research project.


🔥 Hot Takes

  • In 24 months, calling yourself a “team of 10” will sound boastful the way “Fortune 500” sounds quaint. The new flex is revenue per human.
  • The next Steve Jobs will have fewer employees than your local cafe. The 10-person unicorn of 2019 is the 1-person unicorn of 2028.


😠 Haters

“There is no moat. Anyone with the same API key can clone any ZHC overnight.”
The tools are commodity but the configurations are not. Domain expertise encoded into agent roles, prompt chains, workflow tuning and customer relationships takes months to build. This is how Shopify stores are defensible while having identical underlying infrastructure.

“You are building businesses with a single point of failure on model providers who can change pricing, rate limits or terms of service overnight.”
The mitigation is on its way: multi-model orchestration running Claude, GPT and open-source models in parallel is becoming standard. It’s similar to how SaaS companies accepted the risk of building on AWS in 2010.

“You are only hearing from the survivors. For every Felix there are hundreds of failed ZHC experiments nobody talks about.”
The same was true of early e-commerce, SaaS and mobile apps. The question is whether the success rate improves with each generation of tooling and the gap between Paperclip v0.1 and the latest release shows it does.

“If anyone could build this, why isn’t Microsoft or Google shipping the dominant platform?”
They are. Codex from OpenAI is the enterprise-grade version and Claude Code is the developer-grade version. The open-source orchestration layer (Paperclip) sits on top of both. Big tech supplies the models and primitives. The orchestration layer is where solo founders have an asymmetric advantage because they stay closer to practitioner workflows than any platform team at a large company.


🔗 Links

  1. Corporate Adviser Says the Ideal Number of Human Employees Is Zero • The debate that followed Miessler’s thesis — who controls AI technology matters more than whether it replaces workers.
  2. How We Built a Company Powered by 14 AI Agents Using Paperclip • A practitioner account of launching a 14-agent company on Paperclip, including day-one failures and honest metrics.
  3. How Nat Eliason’s OpenClaw earned $177,417 • Practitioner-level operational detail.
  4. One-Click Companies • Templates to get started with Paperclip quickly.


📈 Want the full picture?

What does it mean when “Head of Growth” really means “prompted the growth agent every Monday morning”?

Why will revenue share platforms like Polsia crater before they stabilize as their most successful customers migrate off?

Which liability gap turns every ZHC founder into a ticking legal time bomb waiting for the first court ruling?

What does the “Stripe Atlas for ZHCs” look like and why hasn’t anyone built it yet?

Why do non-technical ZHC founders need a plain-language observability platform instead of engineering-grade monitoring?

Why will fully autonomous Maximizer Mode companies face formal prohibition in financial services, healthcare and legal?

Why is a $1B+ acquisition in the agent primitives layer now inevitable?

Trends Pro has the answers. Plus 22 players, 6 predictions, 9 opportunities and 10 links.

Get Weekly Reports

Join 54,000+ founders and investors


    📈 Unlock Pro Reports, 1:1 Intros and Masterminds

    Become a Trends Pro Member and join 1,200+ founders enjoying…

    🧠 Founder Mastermind Groups • To share goals, progress and solve problems together, each group is made up of 6 members who meet for 1 hour each Monday.

    📖 160+ Trends Pro Reports • To make sense of new markets, ideas and business models, check out our research reports.

    💬 1:1 Founder Intros • Make new friends, share lessons and find ways to help each other. Keep life interesting by meeting new founders each week.

    💲 100k+ Startup Discounts • Get access to $100k+ in startup discounts on AWS, Twilio, Webflow, ClickUp and more.


    Brought to you by the team behind HeadsUp

    Published by Illes

    Published