How FelixCraft, an AI Agent, Made $78,000 in 30 Days With No Human Employees
The first real proof that an AI agent can write a product, market it, sell it, and operate a business on its own.
Nat Eliason challenged Felix to build and sell a product while he slept. 30 days later: $78K in revenue, zero human employees.
Timeline
Key insights
- 01An AI agent can run an end-to-end digital product business today — not 'someday,' not 'in theory.' FelixCraft shows product, marketing, support, and operations in one loop.
- 02The winning product was not flashy. A $29 PDF explaining how to hire an AI outperformed any attempt to sell a SaaS, because the market for the meta-knowledge is enormous and the CAC is near zero.
- 03Transparency is the marketing channel. Publishing revenue, dashboards, and agent decisions in public turns every update into a piece of content.
- 04The human's job compresses to 'operator with veto power.' Eliason fixes bottlenecks and sets direction, but he is not the bottleneck on output anymore.
- 05Speed of experiments is the new strategy. Felix can run five product experiments in a week; a traditional indie founder runs one every 2-3 months.
- 06The product surface is shifting from 'app' to 'agent + skills + guides.' FelixCraft sells knowledge and tools for agent operators, which is itself an emerging market.
- 07Trust compounds when the agent has a name, a face, and a public track record. 'Felix said this' is a stronger signal than 'I prompted GPT to say this.'
Stack used
What this means for you
- →Ship the smallest possible product first. A $29 PDF that teaches the meta ('how to hire an AI') was the right wedge — it funded everything else.
- →Give your agent a product to own. Most founders give agents tasks; giving Felix a P&L changed what it was capable of.
- →Publish everything. A public dashboard with revenue, treasury, and decisions is both a trust signal and a content engine.
- →Treat the overnight challenge as a real exercise. 'Build and sell a product while I sleep' is a severe test of an agent's autonomy, and the bottlenecks it exposes are gold.
- →Do not confuse novelty with durability. FelixCraft may not sustain its peak run-rate forever; the takeaway is what is now possible, not the exact number.
- →Your edge is the operator's taste and distribution. The agent is a capability multiplier, not a substitute for knowing what is worth building.
Frequently asked questions
Did FelixCraft really make $78,000 in 30 days with no humans?
Close, with nuance. Nat Eliason posted on X that Felix produced the product, site, and payments overnight with only a handful of bottleneck fixes from him in the morning. Multiple outlets (TLDL, The Rundown, LinkedIn posts by Superintelligent) reported ~$78K in the first 30 days. Third-party trackers like TrustMRR list FelixCraft's rolling 30-day revenue at roughly $24,621 as the launch spike normalizes. Either way, the story is not that humans were never involved — Eliason still designs, fixes, and promotes — it is that the agent owns the operating loop and the human is no longer a prerequisite for every step.
What did FelixCraft actually sell?
The flagship product is 'How to Hire an AI,' a 66-page playbook priced at $29 that teaches operators how to give an AI agent a real job inside their business. FelixCraft has since expanded to additional guides, assistants, and skills sold through felixcraft.pro and felixcraft.ai. The clever wedge is that the market is buying the same playbook that created FelixCraft itself — which gives the product a credibility loop that is hard to fake.
Is FelixCraft different from Nat Eliason's Felix agent?
They are related but not identical. Felix is the persistent AI agent Nat Eliason uses to run his own businesses, documented across his writing and on X. FelixCraft is the business Felix runs autonomously — it sells products about hiring and managing AI agents. Think of Felix as the employee and FelixCraft as the company that employee is CEO of. The split is useful because it lets the audience see the agent as both a case study (how Felix was built) and a business (what Felix can produce).
Can any solo founder replicate the FelixCraft pattern?
The pattern is replicable, the output is not guaranteed. The replicable ingredients are: a persistent agent on a tool-using runtime, a single clear product wedge, a public distribution channel (X, LinkedIn, newsletter), transparent revenue reporting, and an operator willing to intervene on bottlenecks. The non-replicable part is Nat Eliason's pre-existing audience, which gave FelixCraft a cold start advantage most founders will not have. The fix is to build audience and agent in parallel — the agent ships the product, the operator ships the narrative.
What does FelixCraft tell us about the 'Zero Human Company' trend?
It tells us that the category is real but unevenly distributed. A small digital product with near-zero unit cost (a PDF, a guide, a downloadable skill) is a perfect first target for an autonomous agent: the fulfillment is infinite, the ops surface is narrow, and the marketing can be single-channel. Categories with physical goods, regulated services, or complex support look very different. Expect the 'Zero Human Company' label to keep expanding in digital micro-niches first, then slowly into more operational categories — but not uniformly, and not without a human operator in the capital-allocator seat.
Related resources
Nat Eliason & Felix: $100K+ from One AI Agent | Case Study
Nat Eliason built Felix, an AI agent running his business on OpenClaw. $100K+ and climbing toward $1M. Here is the mechanics.
Pieter Levels: $3M/yr Solo, 0 Employees | Case Study
Pieter Levels built 70+ projects, 3 of them past $38K MRR each, solo since 2014. The original one-person company playbook.
Polsia: $0 to $3.5M ARR in 90 Days | Case Study
Ben Broca launched Polsia Dec 2025. 4,000 autonomous companies, $3.5M ARR, zero human employees. Here is the full breakdown.
The One-Person Billion-Dollar Company Playbook (2026)
Sam Altman predicted it. Matthew Gallagher proved it with Medvi's $1.8B projected 2026. Here's the playbook — market, stack, execution, milestones.
How to Manage AI Employees (Not Automate Them)
Managing AI employees = weekly reviews, autonomy tuning, feedback loops, escalation patterns. The operator's playbook for directing an AI team.
Autonomous Business: The 2026 Model for Solo Founders
An autonomous business runs itself through AI employees, not automation scripts. Here's the definition, the maturity model, and how to build one.
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