I'm a commercial pilot (ASEL, AMEL, Instrument) with 393.1 total hours. My day job is managing a small business in Central California. I'm not flying full time yet. Five hundred hours is my next personal planning threshold, when I will recheck operator openings and requirements; it is not a universal Part 135 minimum or a promise of employment. The later benchmark is 1,500 hours, subject to the applicable ATP pathway.
So I'm building income streams on the side, and using AI automation to do most of the heavy lifting while I'm at work. Here's an honest accounting of where things stand.
The Stack: Hermes Agent + Talos
I originally ran this operating system through OpenClaw. That runtime is now retired. The current stack uses Hermes Agent, with my agent Talos handling bounded research, local builds, verification, and scheduled monitoring on my Mac.
- Content creation: Blog posts, SEO optimization, marketing copy
- Code generation: Helps build and maintain my iOS apps
- Market research: Scans for opportunities, analyzes competitors
- Evidence management: Separates shipped proof, drafts, paper results, and real revenue
- Scheduled monitoring: Runs a bounded set of health, evidence, and briefing jobs
Talos isn't magic. It's infrastructure. The value comes from having a system that moves projects forward during the hours I can't.
What I'm Building (Honest Status)
KDP Books — Live, First Revenue
The aviation catalog now has five Amazon workbooks, with PPL and Instrument carrying the clearest current sales signal:
- Private Pilot License Workbook: 500+ practice questions, $19.99 print / $9.99 Kindle
- Instrument Rating Workbook: Comprehensive IFR prep, $24.99 print / $9.99 Kindle
- The Complete Commercial Pilot Workbook: written test and oral exam prep for commercial candidates
- Multi-Engine Add-On Rating Workbook: PA-44-focused V-speeds, Vmc, OEI, and oral prep
- Clearance to Confidence: VFR Radio Calls Workbook: scenario-based call-building and readback practice
All five are now collected on the Aviation Pilot Workbook Series page so the book funnel has one clean destination instead of scattered product links.
Reconciled tracking at August month-end puts the aviation workbook lane at 149 units, $1,282.32 USD plus €71.47 EUR and A$9.49 in dashboard-estimated royalties. August aviation proof is 27 paid units, 1,187 KENP, and 0 refunds. Full-catalog tracking is 157 books. These royalty figures are not settled payouts. That is tracked aviation-product proof across the stated March-August evidence windows, not an Amazon Ads attribution claim. Still small, but no longer imaginary.
FeedFare — Live on the App Store
An iOS app where you trade steps for screen time. Set a daily walking goal, earn minutes, and use movement to build better phone habits.
FeedFare is live on the App Store and remains a separate product experiment. I do not have verified revenue proof for it. Further conversion and App Store work is parked unless the lane is explicitly reopened.
Kalshi Weather Bot — Legacy Hypothesis Closed
The retired credential-free weather stack collected public Kalshi weather-market, NBM, and METAR evidence. It was a paper/shadow research system, not a live trading product.
An earlier paper headline failed harder market-relative validation. The settled V1 ledger finished 1 win, 19 losses, and -$24.99 in realized paper P&L. V2 later lost to the market baseline or found zero opportunity density, V3 could not produce a trustworthy leakage-safe holdout, and V4 found zero valid settlement-source crossings or executable transitions. The weather hypotheses are killed, collectors are paused, and live trading is off. The useful lesson is that automation still needs brutal gates.
Other Apps — Awaiting Release or Parked
Inflection Point is awaiting App Store release, with a planned 3-day access preview, free plan, and optional $14.99 lifetime unlock. It is not a verified revenue lane. Older experiments such as SpendSlap and PetFace remain parked; unfinished software is not a product merely because AI generated code for it.
YouTube — Not Yet Monetized
My YouTube channel documents the journey. AI helps with content planning, script drafts, and thumbnail concepts. It's not generating income yet. I'm treating it as a long-term play and a transparency tool for the Flight Funded experiment.
What AI Actually Does Well
After several months of running this system, here's what genuinely works:
- Bounded background work: Talos monitors evidence, checks system health, and prepares concise decision briefs without pretending activity is progress.
- Research at scale: Competitive analysis, niche scouting for KDP, SEO keyword research. Tasks that would take me hours take minutes.
- Code generation: Going from idea to working iOS app in days instead of months. I'm not a developer by training. AI closed that gap.
- Content production: Blog posts, product descriptions, social media drafts. The volume would be impossible solo.
What AI Doesn't Do
- Ship products by itself: Public releases, account changes, purchases, and outreach still require human judgment and approval.
- Build an audience: No shortcut for genuine engagement. Reviews, followers, and trust take time.
- Replace judgment: The legacy weather hypotheses were closed when the evidence failed; blindly trusting model signals with real money is how you go broke.
- Guarantee income: KDP royalties are real. Paper P&L, app code, clicks, and unfinished products are not revenue. Still early, no fairy dust.
The Math
The near-term personal planning threshold is 500 hours. At 393.1 hours, that leaves 106.9 hours. At roughly $200 per hour, that is about a $21,380 gap. Reaching 500 will trigger a new market check against real operator, insurance, aircraft, route, and medical requirements; it does not create automatic eligibility.
The cleanest signal right now is aviation products: reconciled tracking at August month-end is 149 aviation units and $1,282.32 USD plus €71.47 EUR and A$9.49 in dashboard-estimated royalties, not settled payouts. Full-catalog tracking is 157 books. The workbook catalog is the lead revenue lane. FeedFare is live but unproven and parked. The legacy weather hypotheses are closed, their collectors paused, and live trading off. Any newer research is separate, unproven, and not a funding contribution or live-readiness claim; WeatherNext remains parked.
What's Next
- Sell what is already shipped: The aviation workbooks need distribution and customer proof more than another product draft.
- Hold ad scale: Keep the converting Instrument campaign capped while same-window royalty economics mature.
- Pursue paid flying honestly: Apply only where the current posting and verified logbook fit.
- Preserve the weather closure: No collector or live restart for the killed legacy hypotheses. Separate research does not reopen them or establish income.
- Leave parked products parked: No app resurrection without a buyer, user, or explicit proof gate.
I'm tracking every dollar publicly on my Flight Funded dashboard. Real numbers, real time, no spin. If AI-powered side income can fund a pilot career, you'll see it happen there first.
The honest version of this story isn't "AI made me rich." It's "AI gave me leverage I didn't have before, and now I'm trying to turn that leverage into something real." I'll let you know how it goes.