I Wrote a Playbook for Running OpenClaw Without Writing Code

Six months ago I couldn't tell you the difference between JSON and JavaScript. I'm a commercial pilot who manages a car electronics shop in Merced, California. I'm 46. I have two adult sons and a mortgage. I am not the target demographic for "AI agent frameworks."

But I needed more flight hours. And flight hours cost money. So I started building things that make money while I'm at work, and I used OpenClaw to do it.

What I Actually Built

My OpenClaw agent runs on an iMac in my living room. Here's what it does while I'm at work selling car stereos:

• Trades weather prediction markets on Kalshi. 66 live trades, 57.9% win rate, +$279 realized profit. The bot scans temperature forecasts, finds mispriced brackets, and places trades with position sizing based on Kelly criterion.
• Monitors my iOS apps on the App Store. Checks review status every few hours, alerts me on Telegram if anything gets approved or rejected.
• Generates a daily intelligence briefing. Scans RSS feeds, runs Tavily searches, synthesizes with Gemini, and posts to my Discord and Telegram before I wake up.
• Backs up everything. Workspace to GitHub, database to Synology NAS, memory files nightly.
• Tracks revenue across KDP, Gumroad, and Kalshi. Weekly rollup sent to my Telegram DM every Sunday.

None of this required me to write code. I described what I wanted in plain English, and OpenClaw's coding agents built it. When something broke, I described the error and it got fixed.

The Mistakes That Cost Me Money

The playbook isn't a manual. It's a case study with every disaster included.

Like the night I accidentally spawned 23 sub-agents at full Opus pricing and burned through $340 in API credits before I noticed. Or the week I ran all 47 cron jobs on the same AI provider and hit rate limits so hard that my trading bot stopped placing trades for 12 hours during a profitable window.

Or the three apps I built, submitted to the App Store, and then realized weren't solving real problems. Those are documented too, with the exact moment I should have killed each one.

Who This Is For

You don't need to be a pilot. You need to be someone who wants an AI agent doing real work, not just answering questions. If you have a Mac, a Telegram account, and a willingness to pay $30/month in API costs, you can run what I run.

The playbook is a practical starter kit built from real use: copy-paste cron configs, reusable prompt templates, and the mistakes that taught me where agent systems actually break. Not screenshots of menus. Not theory.

Inside the Playbook

Here are three real pages from the PDF so you can see what you're actually buying: cover, table of contents, and an interior page from the security chapter with concrete config examples.

Cover. The actual PDF buyers get, not a mockup pretending to be a book.

Table of contents. Fifteen chapters covering setup, memory, cron jobs, dashboards, security, sub-agents, and monetization.

Interior page. Real operating detail, including config patterns, security rules, and the mistakes that keep this stuff from getting expensive.

Start with the 16-page free sample, or get the full PDF + EPUB. Use code FIRSTFLIGHT for 30% off.

The Bigger Picture

All of this feeds into Flight Funded, my public experiment to fund flight hours through side projects. The playbook is one revenue stream alongside KDP workbooks, a Kalshi trading bot, and iOS apps. Every dollar tracked in public.

I need 131 more hours to qualify for Part 135 operations. At ~$200/hour for rental and instruction, that's $26,000. The bot is averaging $40/day. The math is starting to work.