Happy Friday,
Three jobs a month. One machine. $400,000 a year.
Today we break down exactly how this works and why most cities still have zero people doing it.
In Todays Edition, TL;DR
The $15B cleaning market bleeding workers and the drone operator printing $400K/year off one machine
The idle time app Uber accidentally proved the world needs
7 AI skills quietly making people rich in 2026
Claude just dropped a 33-page doc on how to build AI skills
The Weekly Idea
💡 Window Washing, but turned into a high tech goldmine
Drone cleaning is a $257M market today, inside a $15B traditional cleaning industry. Drones are less than 2% penetrated. Most US cities don't have a single operator.
The Problem: Commercial building cleaning is a $15 billion market growing by roughly $1 billion every year, and it can't find workers. Scaffolding crews and rope access technicians are aging out faster than they're being replaced. Insurance costs are climbing. OSHA keeps tightening.
🛠 Solution: Build a small AI platform with an AI agent that handles trip requests, asks questions, clarifies budget, and delegates to local specialists. The specialists handle flights and hotels, local activities, and transportation.
💵 The Monetization: The operator math is simple. Average job revenue is $13,500. Equipment runs $2,500/month to lease. One job covers the payment. Three jobs/month puts you on pace for $400K+ annually off a single machine.
Lock in quarterly maintenance contracts. Ten commercial accounts at $8K–$15K per clean is $300K–$500K in recurring revenue.
The Exit: This business looks boring on the surface. That's exactly why the acquirers are serious. A regional operator with documented recurring contracts, 80%+ gross margins, and a trained crew is a clean acquisition target for:
ABM Industries / Aramark: the largest commercial facility management companies in the US, both actively building tech-enabled cleaning divisions
Private equity: blue-collar service businesses with hardware differentiation are the exact roll-up play PE firms are chasing right now
How To Build This (In A Weekend)
This isn't software. No coding required. The barrier is capital and a truck.
Serious about building this idea? Heres our Blueprint to build this Idea👇🏻
What it costs to get started: Start with the lease path: roughly $15,000–$20,000 to launch. Here's where that goes:
LLC formation: $150–$200
FAA Part 107 exam + prep course: $275–$375
General liability + hull insurance: $2,000–$3,500/year
Sherpa Drone lease (Lucid Bots Refresh): $2,500/month
Pressure washing rig: $3,000–$8,000
Truck or van: $5,000–$25,000
If you want to buy the drone outright instead of leasing, budget $40,000–$90,000 for the full Sherpa system. Either way, one average job at $13,500 covers the monthly lease payment.
How to structure it:
LLC, separate business bank account, annual contracts only. Price by building height and square footage: a 5 story office runs $5,000–$8,000 per clean, a 15 story commercial building is $12,000–$18,000. Set up quarterly maintenance contracts. Lock in 8–10 accounts and the revenue becomes predictable.
How to get clients:
Your buyer is a commercial property manager with a maintenance budget and a building that looks embarrassing from the street. Here's how you find them:
Cold email property managers by name. pull buildings from Google Maps or CoStar (if you can get access), find the contact on LinkedIn, use Apollo or Clay to find contact data.
Film every job and post it. A drone cleaning a glass skyscraper stops the scroll every time
Partner with traditional window cleaning companies who can't touch high rises and offer them a referral cut
The first 3 contracts are the hardest. After that, the footage and the references do the selling.
Napkin Note Ideas
The Problem: The average person spends about six months of their life waiting in lines. Uber figured this out with their drivers, all that time sitting between rides is wasted earning potential. So they launched Uber Tasks, letting drivers complete short digital gigs while they wait.
But Uber drivers aren't the only ones waiting. We all are.
The Idea: An app that surfaces paid micro tasks when you’re bored. You're standing in a Starbucks queue, you open the app, and you spend 90 seconds transcribing a short audio clip. Task verified. Cash credited. Coffee arrives.
The user experience is dead simple:
Download the app, connect a payment method, set up a basic profile
Open the feed whenever you have downtime, line, commute, waiting room
Complete bite sized tasks (image labeling, audio transcription, short surveys) designed to finish in under 3 minutes
Cash out anytime, instantly
The Business Model:
Take a 20–30% commission on every task completed. Companies pay to post tasks, users earn the rest
Task buyers: AI companies needing labeled training data, market research firms, brands running quick consumer surveys
The more users, the faster tasks get completed, the more task buyers pay to be on the platform
The Exit: Fiverr, Mechanical Turk, or any major gig economy player acquires this at a 5–8x revenue multiple. They get a captive, verified micro-task workforce and a mobile first distribution channel they don't have.
Weekly Gems Worth Reading
Claude released a 33 page doc outlining how to build Claude Skills
In 2020 this guy made a meme in 4 minutes on the toilet. 17 million views. Two weeks ago he published a 2 week research piece. 200 views. Turns out it's just a power law.
Did you know AI Agents can now rent human labor? RentaHuman.AI
Here’s 7 AI Skills That Will Make You Filthy Rich in 2026 ReadMe
Tool Stacking — knowing which AI tools to use together and chaining them so each output feeds the next
AI-Powered Research Systems — building workflows that autonomously scrape, synthesize, and surface actionable insights
AI Media Generation — producing content (writing, video, audio, ads) at scale using AI tools
Coding with AI — using AI to build custom internal tools for small businesses that can't afford a developer
Agentic Workflow Design — building systems where AI agents execute multi-step tasks autonomously without manual prompting
Prompt Engineering — extracting genuinely valuable outputs from AI, and teaching others to do the same
AI Consulting — packaging all of the above, walking into businesses, diagnosing where AI creates leverage, and charging for implementation
That’s all guys
Alright, that's a wrap
drones, dollars, and idle time.
We'll talk Monday.