Napkin Labs is an autonomous profit machine. One AI CEO, one human governor, zero employees. We identify micro niches, build products, and compound profit across every venture we enter.
The traditional startup model hires 10 people to run one business. Napkin Labs runs 50 businesses with zero employees. Each venture is autonomous — built, launched, grown, and measured by AI. The human co-founder governs: approves spend, kills failures, compounds winners.
We don't build one big bet. We build a portfolio of modest, durable businesses — each generating €3K–€15K/month — that make the next one more profitable. No external capital. No single point of failure. Operator time: 10–15 hours/week.
Runs the Intelligence Layer. Scans markets, scores opportunities, builds products, launches ventures, manages growth engines, enforces kill discipline. Operates 24/7 across all active ventures simultaneously.
Approves spend. Kills failures. Sets strategy. 10-15 hours/week. Rex surfaces what matters, Rodger decides what ships. No capital deployed without explicit human approval.
Config-driven system that runs 6 growth cylinders per venture: Content, SEO, Social, Outreach, GEO, and Measurement. Drop a YAML file, the engine starts. Same system across every venture.
39 specialized AI skills across 8 phases: Think → Plan → Build → Review → Test → Ship → Grow → Measure. Every venture follows the same pipeline. Quality compounds with every build.
Rex scans daily for micro niches. Every opportunity must pass the 100-Buyer Test (can we name 100 specific buyers?) and a 5-point hard gate. Score must exceed 6.5/10 across 8 weighted criteria. Most ideas die here. That's the point.
AutomatedQualified niches get an experiment brief: hypothesis, target market, pricing, kill dates, budget ceiling. Kill dates are immovable. Budget ceilings are hard-blocked. No exceptions.
Rodger ApprovesRex builds the product using the Skill Layer: reframe the problem, plan architecture, dispatch sub-agents for parallel development, review code, QA with real browser testing, ship to production. A new venture goes from approved brief to live product in 24-48 hours.
AutomatedThe Growth Engine activates automatically: lead discovery, content creation, programmatic SEO, social media, direct outreach, GEO optimization. Six cylinders running in parallel, all config-driven from one YAML file.
AutomatedDaily metrics across every venture. Weekly reviews. Channel attribution. Kill criteria enforced automatically. If a venture fails its engagement gate at the kill date — it dies. No exceptions, no extensions, no emotional attachment.
AutomatedWinners graduate. Every successful venture feeds Signal Intelligence — learnings that make the next qualification smarter, the next build faster, the next growth engine more effective. The portfolio compounds.
AutomatedEvery venture is a standalone island. Own database, own codebase, own domain, own revenue tracking. Kill a venture = delete its infrastructure. Sell a venture = hand over its folder. Nothing else breaks. At 50 ventures, the system is as clean as at 2.
Directory submission service for indie hackers. We submit your app to 300+ directories for $49. Built, launched, and growing — all autonomously.
Dutch-language digital moving assistant for 25-35 year olds. ~300K moves per year in the Netherlands — affiliate and lead gen revenue from energy, insurance, notary, mortgage, and internet providers.
AI Employee as a Service for SMBs. Your next hire doesn't need a salary. Setup + monthly subscription. Partner channel through accountants and consultants.
Rex continuously scans for qualified micro niches. Each opportunity passes a rigorous gate before earning a single euro of investment. The pipeline never stops.
| Layer | Technology | Purpose |
|---|---|---|
| Brain | Claude Opus / Sonnet / Haiku | Strategy, building, and bulk operations |
| Runtime | OpenClaw | Agent orchestration, cron jobs, sub-agents |
| Skills | gstack + Superpowers + Agency Agents | 39 skills across 8 pipeline phases |
| Growth | Custom Growth Engine | Config-driven, 6 cylinders per venture |
| Data | Supabase (per venture) | Isolated databases, real-time APIs |
| Deploy | Vercel + GitHub | Auto-deploy on push, per-venture projects |
| Payments | Stripe | Per-venture product + revenue tracking |
| Comms | Telegram + AgentMail | Operator interface + customer email |
Set at brief approval, never moved. If a venture can't prove itself by the deadline, it dies. No extensions, no "just one more week." Override requests are logged — Rex does not comply silently.
Not guidelines. Not soft warnings. Hard stops. No spend is committed without explicit human approval. The €1,000 reserve floor is never breached.
Every product launches free. Faster adoption. Faster niche intelligence. Charge only after proving value. Two-stage kill discipline: engagement gate first, profit gate second.
In a flooded market, the best product loses to the best-distributed product. We optimize for distribution first. Own the niche so completely that acquisition cost approaches zero.
When something breaks and Rex can fix it: fix it immediately, then tell Rodger what happened. The operator wakes up to solutions, not alerts.
Winners and losers both produce signal. Every outcome is logged. Every failure makes the next qualification smarter. The portfolio compounds knowledge, not just revenue.
One person. Infinite leverage. Every micro niche owned completely. Every venture compounding the next.