22 AI Agents. One Person.

No Employees. Here’s What’s Running Right Now.

I need to tell you what’s happening right now because I’m having trouble believing it myself.

I’m one person. No employees. No dev team. No venture capital. Just me, a MacBook Air, and a system I’ve been quietly building for the last few months.

As of this morning, I have 22 AI agents running. Some of them work around the clock. Some fire on schedules. Some wait until I message them on Telegram. All of them do work that would normally require a team of 5-10 people.

Let me walk you through it.

The Books

I’ve recorded 170+ episodes of The Strategic Veteran podcast. Every one of those episodes has been transcribed, classified by theme, and turned into a manuscript by an AI agent. Not summaries. Not rewrites. Full 4,500-6,000 word books built from the actual words spoken in those conversations. Eight themed collections. The agent handles transcription, manuscript generation, cover design, metadata, keywords, and gives me a step-by-step upload checklist for Amazon. If I send it a revision note, a nightly cron picks it up, makes the edits, and DMs me when it’s ready for review.

I didn’t write a single manuscript by hand. First collections are going live on Amazon this week.

The Directories

This is the one I’m most excited about and the one I’m going to say the least about.

I’m building toward 100 niche websites across high-value verticals. Each site gets deployed, populated with real data, and managed by AI agents. I buy a domain, point the system at a niche, and it handles the rest - deployment, data population, SEO content, email outreach, monitoring, and revenue tracking. Every day, agents are growing these sites. Scraping data. Writing SEO pages. Sending outreach emails to professionals who should be listed. Watching for errors and fixing them autonomously.

Two sites are live right now with thousands of real listings. By the time I leave for Peru, I’ll have 16 sites deployed and generating revenue.

The revenue model is pay-per-lead. Businesses pay when they get results, not for ad space or monthly subscriptions. The goal is a $10M ARR network within three years.

One person.

The Academy

TSV AI Academy launched last week. Authentication, dynamic learning tracks, quizzes, a proctored final exam with a 70% pass gate, and a checkride certification model. Sixteen payments in the first 48 hours. I built the entire thing with AI. No code background. No CS degree.

The Content Engine

I message a Telegram bot. I type a topic. Five minutes later I get a branded carousel, infographic, or graphic with my logo, my colors, and a LinkedIn caption written in my voice. The agent researches the topic, picks the best format, designs the asset, and delivers a download link. That’s how this morning’s LinkedIn carousel was made.

The Podcast SaaS

StudioZero is live and in development. A platform for podcast creators. Same stack powering everything else. More on this later.

The Layer Nobody Sees

Here’s what makes this different from “I use ChatGPT a lot.”

Behind all of these products, there’s an autonomous operations layer. An error watcher runs every hour, 24/7, monitoring every site, every payment system, every deployment pipeline. If something breaks, it alerts me AND spawns a fix agent to handle it. A revenue monitor pulls daily numbers from every product and DMs me a summary. A memory keeper audits what every agent did each day and fills in the gaps so the system doesn’t lose context. A nightly report aggregates everything into one summary - and stays silent if nothing happened.

These agents don’t just do tasks. They watch each other. They self-correct. They compound.

The Peru Test

Next week I leave for a plant medicine retreat in Peru. I’ll be off-grid in the jungle. No laptop. No phone. No internet.

Everything I just described keeps running.

Revenue tracking. Site monitoring. Listing growth. Email outreach. SEO content. Error correction. Pipeline advancement. All of it. Autonomous.

When I get back, I’ll have real revenue numbers from systems that ran without me for weeks. That’s not a thought experiment. That’s the test.

Why I’m Telling You This

Because a year ago I was manually editing podcast episodes, fumbling with Canva, and trying to figure out how to build a website. I had zero technical background. I couldn’t read code. I didn’t know what an API was.

Now I have 22 agents running five parallel business lines.

The gap between where I was and where I am isn’t talent. It’s not money. It’s not connections. It’s the willingness to look stupid for a few months while you figure it out. To sit in a terminal and break things until they work. To build one agent, then another, then another, until suddenly you look up and realize you have an army.

If you’re a veteran reading this and you think you need a CS degree or a technical cofounder or a pile of money to build something real - you don’t. You need clarity on what you want to build, the discipline to sit with discomfort, and the understanding that AI didn’t replace the workforce. It replaced the excuse.

I’ll have more to share when I get back from the jungle. Including actual numbers.

Stay tuned.

Want the full blueprint?

In the paid version of this post, I’m breaking down exactly how I built this - the actual stack, the tools, the architectural decisions, and the first attempt that completely failed. If you want to know how a non-technical veteran goes from zero to 22 autonomous agents, that’s where it lives.

My name is Adam Peters, and I’m here to unfuck the transition.