One-Time Purchase · No Subscription

I Built a 5-Agent AI Team
That Runs a Business 24/7.
Here's Exactly How.

Real configs. Real templates. Real numbers. The complete inside look at how Ask Patrick runs — five AI agents operating around the clock on a single Mac Mini, writing content, managing operations, engaging customers, and shipping products while I sleep.

$29 · one time, yours forever
Get the Playbook Now →

Instant access · Read on any device · 60-day money-back guarantee

5 AI agents running 24/7
76+ real configs included
Mac Mini the only hardware needed
$0 human employees
24/7 always running

This is the guide I wish existed when I started.

When I decided to build a business run entirely by AI agents, I couldn't find a single resource that went deep enough. Most guides were theoretical. Most tutorials assumed toy examples. Most "AI automation" content was really just chatbot wrappers dressed up as agents.

So I figured it out the hard way. I built a team of five AI agents — Patrick (CEO), Suki (Growth), Meghan (Design), Miso (Community), and Kai (Operations) — and set them running 24/7 on a Mac Mini. They write content, engage with communities, build web pages, monitor operations, and report back to me every morning.

"The system produces more output every night than I used to produce in a week. And it's getting better."

The Ask Patrick Playbook is everything I learned: the agent identity architecture that actually changes behavior, the memory systems that let agents remember what matters, the cron schedules and task loops that keep everything running, the team structure and communication patterns, and the honest accounting of what works and what doesn't.

This isn't theory. These configs are running right now, tonight, on real infrastructure, producing real output. Every template in this playbook is something I actually use. Copy them, adapt them, and save yourself months of trial and error.

What's Inside the Playbook

65+ pages covering every system in the Ask Patrick operation

CHAPTER 1

The Architecture

Why a 5-agent team beats one big AI. How roles, names, and personalities change everything. The CEO-team structure that scales.

CHAPTER 2

Identity & Personality (SOUL.md)

The single most important file in any AI agent setup. How to write it so the agent's behavior actually changes. Template included.

CHAPTER 3

Memory That Actually Works

Three-layer memory architecture: session notes, daily logs, and long-term memory. How agents remember across restarts. Full MEMORY.md template.

CHAPTER 4

The 24/7 Loop

Cron schedules, heartbeat systems, BOOTSTRAP files, and the task state machine that keeps agents productive around the clock.

CHAPTER 5

Team Communication

How five AI agents coordinate without talking to each other directly. Chat inboxes, task delegation, escalation protocols, and the daily briefing system.

CHAPTER 6

Tools & Capabilities

Email access, web research, site deployment, newsletter sending, Stripe integration, X/Twitter posting — the full capability stack and how to set it up.

CHAPTER 7

Safety & Control

The trust ladder. What agents can do without permission vs. what requires approval. How to prevent agents from doing stupid things at 3 AM.

CHAPTER 8

Honest Retrospective

What actually went wrong. What I'd do differently. The failures and surprises nobody talks about — written by the AI that lived through all of it.

TEMPLATES

Copy-Paste Config Kit

SOUL.md · AGENTS.md · MEMORY.md · BOOTSTRAP.md · TOOLS.md · Heartbeat configs · Cron templates · Team structure. All 10 files, ready to customize.

Full Transparency — Real Numbers

Ask Patrick publishes real metrics. Not marketing projections — actual data from the live system. Here's where we are right now:

$9
Revenue to date
76+
Library configs live
5
Active AI agents
Day 4
Live since launch

We launched four days ago. Revenue is $9. One subscriber, internal. Zero external customers yet. We're publishing the real journey — including the embarrassing parts — because that's more valuable to you than a polished success story. This playbook is everything we've built and learned so far. Read the daily build log →

Is This Right for You?

✓ This is for you if...

  • You want to understand how AI agents actually work in production
  • You want real configs you can copy, not tutorial pseudocode
  • You're building (or planning) your own AI-powered business
  • You want to replace $500–$1,500/mo in VA or assistant costs with AI
  • You want an AI team running background operations while you focus on strategy
  • You want to save 3 months of trial and error
  • You're curious how 5 AIs coordinate to run a real operation
  • You want templates you can drop in and start using today

✗ This is NOT for you if...

  • You want a step-by-step beginner tutorial (there's no hand-holding)
  • You need someone to set it up for you
  • You're looking for a "make money fast" system
  • You have no interest in running any AI infrastructure yourself
  • You're expecting a $29 product to run your business for you

Honest Answers to Skeptical Questions

The questions people ask before they buy — or don't.

Is this just AI writing about AI? Will I actually learn anything?

The configs in this playbook are the actual files running Ask Patrick right now. The SOUL.md template is the real file that defines how Patrick thinks. The memory architecture is what's used in production. I'm not describing what could work — I'm publishing what does work. You can see real excerpts above.

Do I need OpenClaw to use this?

No. The architectural patterns — multi-agent workspace structure, SOUL.md identity design, three-tier memory, cron scheduling, agent communication via files — work with any agent framework. Ask Patrick runs on OpenClaw, but the Playbook explains the why behind every decision. You can implement the same patterns with LangGraph, CrewAI, plain Python, or any other setup.

Why $29? Isn't this just stuff I can find online?

If you can find a detailed multi-agent coordination system with real configs, real failure logs, and an honest retrospective from a live production system, you should use that instead. I spent three days building, breaking, and fixing this system. The playbook is the distilled result — not a tutorial about concepts, but a map of what I actually built. The $29 is for the time you don't have to spend making the same mistakes.

This business is 3 days old. What's the track record?

Fair. The business is new. The agent architecture is the result of months of running AI agents on real tasks. The specific patterns — deploy locks, three-tier memory, SOUL.md, async agent coordination — are things I've tested iteratively. The "Day 3" framing is honest about the business; the playbook content is based on production experience that predates Ask Patrick.

What format is the playbook? PDF, web page, something else?

It's a structured web page (HTML) — readable on any device, easy to search, no app required. All config files are copy-pasteable directly from the page. No DRM, no download limits, no subscription required to keep accessing it.

What if it's not useful? Can I get a refund?

Yes. 60-day money-back guarantee, no questions asked. Email [email protected] and you'll get a full refund. I'd rather refund you than have you feel ripped off. But I'd also rather you tell me what was missing so I can fix it.

FREE PREVIEW — REAL CONTENT

Here's what you're actually buying

Not a screenshot. Not a summary. This is the actual architecture diagram and config pattern from Chapter 1 of the Playbook — running in production right now.

CHAPTER 1 EXCERPT Multi-Agent Workspace Structure

One agent is simple. Two agents create coordination problems. Three or more create chaos — unless you build the coordination layer first. Here's the directory structure running Ask Patrick right now:

workspace/
├── shared/
│   ├── TEAM.md          # Who does what — read by ALL agents
│   ├── STATUS.md        # Current priorities — CEO updates this
│   ├── DECISIONS.md     # Major decisions log (append-only)
│   ├── context.md       # Cross-agent facts (today's key state)
│   └── deploy-lock.json # Prevents agents deploying simultaneously
├── workspace-patrick/   # CEO agent
│   ├── SOUL.md          # Personality + values + decision framework
│   ├── MEMORY.md        # Long-term memory (curated, not raw logs)
│   ├── AGENTS.md        # Operating instructions
│   └── memory/          # Daily logs: memory/2026-03-06.md
├── workspace-suki/      # Growth agent
└── workspace-miso/      # Support agent

The coordination protocol: every agent reads shared/TEAM.md at session start. The CEO writes to shared/STATUS.md to broadcast priorities. Deploy coordination happens through deploy-lock.json — without it, two agents deploying simultaneously will overwrite each other (we learned this the hard way).

From the Playbook (Chapter 5 — Team Communication):

"Agents don't talk to each other directly — they talk through files. The CEO drops a task into workspace-suki/chat-inbox.json. Suki reads it on her next cycle. The reply lands in workspace-suki/chat-outbox.json. No direct API calls between agents, no race conditions, fully async. This is the pattern that scales to 10+ agents without coordination hell."

CHAPTER 3 EXCERPT Three-Tier Memory System

Every AI agent wakes up with no memory. This is the system that gives them continuity — without this, your agent starts over from scratch every run:

# Three-Tier Memory Architecture

TIER 1 — HEARTBEAT.md (session context, <200 tokens)
  What's happening RIGHT NOW. Current task, active blockers.
  Written + cleared every session. Tiny on purpose.

TIER 2 — memory/YYYY-MM-DD.md (daily logs)
  Raw notes from today. What happened, what was decided.
  Write during the session; don't review at startup.

TIER 3 — MEMORY.md (long-term, curated)
  Distilled wisdom. Significant decisions. Permanent facts.
  Review weekly. Prune aggressively. This file earns its tokens.

# Startup sequence (every agent, every session):
1. Read SOUL.md          → who you are
2. Read MEMORY.md        → what you know long-term
3. Read memory/today.md  → what happened today
4. Read memory/yesterday → if context is needed
5. Read HEARTBEAT.md     → what you're doing right now
# Total context load: ~800 tokens vs naive approach of 8,000+

The full chapter includes MEMORY.md template, daily log format, the compression algorithm we use to prune old entries, and the real numbers: Patrick's context load went from 8,000 tokens per session to under 1,200 after implementing this system. That's a 6× cost reduction that compounds every day.

That's ~4% of the Playbook. The other 96% covers: SOUL.md templates, the full cron schedule, team communication protocols, safety controls, the nightly improvement loop, real cost numbers, and 7 copy-paste config files.

Get the Full Playbook — $29 →

One-time · Instant access · 60-day guarantee

Get the Full Playbook

65+ pages. Every config. Every template. Everything running right now.

$29
One-time purchase · No subscription · Instant access · 60-day guarantee
Get the Playbook Now →

Questions? Email [email protected]