About Badmutt

Built it on himself. In six weeks.

Garrett Mastro is a full-time SPX 0DTE options trader who built a 22-strategy systematic trading system.
Then he turned that same checklist-driven methodology on his own AI stack. What follows is exactly what he built and how long it took.

The founder built it first.
On himself. In six weeks.

Garrett Mastro doesn't teach frameworks he read about. He's a full-time options trader who built a fully automated 22-strategy trading system — profitable across every quarter it's run. Then he turned that same systematic, checklist-driven methodology on his own AI stack. What follows is exactly what he built, and how long it took.

Starting point

Six weeks ago, his AI setup was: ChatGPT, used occasionally, with zero memory and zero integration. Every conversation started from scratch. 2.6GB of documents scattered across Google Drive with no system. Newsletter written manually — hours per issue. No monitoring, no automation, no agents. Just a browser tab he opened when he needed something.

Before → After

Feb 16, 2026

  • ChatGPT with no memory — re-explained everything every session
  • 347 documents scattered across Drive, unsearchable
  • Newsletter written manually — hours per issue
  • Zero automated monitoring of markets, news, or security
  • No agents, no system — just individual tools used one at a time

April 3, 2026

  • 5 coordinated AI agents — context preserved across the operating stack
  • All documents organized, indexed, searchable, auto-backed-up
  • Newsletter: screenshot → AI draft → edit → publish in minutes
  • 9 automated jobs running 24/7 — intel, security, health, backups
  • 7-model Board of Directors for independent decision review

By the numbers

6

weeks, start to full system

7

coordinated AI agents running 24/7

9

automated jobs, zero daily intervention

10+

hours/week reclaimed (conservative)

347

documents organized from chaos to searchable

$0

per-token cost for local LLM operations

The build

WEEKS 1–2

Established the AI agent's identity and operating rules. Audited 2.6GB of Google Drive — 347 documents, 360K words — and built a 14-category taxonomy. Set up Drive API integration and moved 102 files into the new structure. Zero to organized in two weeks.

WEEKS 3–4

Ran a failure audit on 1,494 messages to find every breakdown. Deployed Scout (daily intel monitoring across 89 accounts) and Sentinel (twice-daily security sweeps). Built the Board of Directors — 7 AI models giving independent reviews on major decisions. Installed a local LLM for zero-cost classification at 93.1% accuracy. Added local audio transcription, automated health checks, and a monitoring dashboard.

WEEKS 5–6

Migrated everything from a laptop to a dedicated Ubuntu server. Deployed 9 automated cron jobs that run without intervention — intel, security, backups, health checks, and a supervisor that monitors the monitors. Added a knowledge base librarian bot, automated newsletter workflow, local embeddings replacing cloud APIs, and a smoke test suite with 23 automated checks. Built and launched badmutt.com.

This isn't a hypothetical. It's not a demo environment. It's the actual system running the business you're reading about right now — this website was built, reviewed, and deployed by the same AI stack described above. The methodology that built a fully automated trading system with 15 consecutive profitable quarters is the same methodology we apply to your AI workflow in the cohort.

The pack

Maia portrait

Maia is the publishing agent. Sophia is the orchestrator. Occam reads the market. Scout, Sentinel, and the rest of the pack handle their own beats. Mastro runs the house. Anything Badmutt ships came from this stack.