Built, tested, explained

I build supervised AI operating systems for small businesses. In the open.

Here is what I have built, how it actually works, and free resources you can use yourself. No accounts, no pitch. I show the work, not just the claims.

Learn the basics

New free resource — The Follow-Up Leak Map

What you see below is what is real today, not a roadmap.

01Proof of work

The work should be visible before the pitch is.

Three things are real right now. None are paid engagements. They are my own work, built and tested in the open.

demosynthetic / supervised run

A follow-up workflow that drafts, then stops before it sends

One supervised run, end to end. A new lead arrives, the system notices the follow-up is missing, summarizes the context, and writes a reply in front of you. Then it holds. Nothing leaves until a person approves, edits, or rejects it.

what this provesMy automation stops before it acts, so you stay in control.

New lead
Missing follow-up
Context
Draft
Held

New lead · Jordan B. (synthetic sample)

Waiting on the next supervised pass.

READY: press run to watch a supervised pass

Synthetic sample. One supervised workflow. Nothing is ever actually sent.

artifactartifact / offline learning lab

A beginner's AI learning lab I built, then gave away

Screenshot of the AI From Scratch learning lab: a beginner's lab for using AI well, with Start, Track 1, Track 2, Flashcards, Drills, Pass Gates, Prompts and Glossary sections. — AI From Scratch. Runs fully offline: no accounts, nothing sent anywhere.

A self-contained course for a true beginner: a placement check, two tracks (use AI well first, then build small workflows safely), flashcards, drills, and pass-gates you only tick when they are true on a normal day. It runs fully offline, no accounts. I built it to teach someone from zero, then opened it up.

systemdescribed only / no private data

An operating system I run my own work on

Claude Codeorchestration

Obsidianmemory

n8nworkflows

Reportsoperating record

Playwrightverification loop

I do not just recommend this way of working. I operate on it every day. Each piece below has a job, and every build leaves a written record I check before I trust it. Described here, never exposed: no private data, no sensitive screenshots.

The rule under all of it

Held before send.

Every send, charge, or commit waits for a human yes — the rule the demo, the resources, and the method all rest on.

02Free resources

Give-first. The things I would want you to read before spending a dollar.

Plain-English field notes on running a business with AI: a concept, why it matters, and what to look at in your own operation. I add to these as I publish. Filter by what you need, then open one to read it right here.

04The method

How I work. Deliberately boring, because boring is what makes automation safe.

Five rules I do not break. Pick one to see what it means and how it shows up in the actual build.

principle 01

Supervised, not autonomous

Everything runs through human approval before it sends or acts. You stay in control. AI does the repetitive part, with you in the loop. Not AI running your business.

in practiceEvery send, charge, or commit waits for a human yes.

05About

I learned how businesses actually run before I automated anything.

My path runs from software into operations and back. I started close to code, then went deep on marketing and lead generation for local service businesses, where the real outcome is never just more leads. It is booked jobs, faster follow-up, and revenue you can predict.

That is why I care about the operating layer underneath the offer: the handoffs, the data, the follow-up, the reporting, and the places work gets lost.

Software orientation

Computer science track at the University of Tampa, plus early project work from co-founding a student tech club.

Marketing into systems

Lead generation pulled me past ads into attribution, qualification, speed-to-lead, CRM logic, n8n, and Airtable.

AI as the operating layer

In early 2026 I moved from using AI as a tool to building a masterclass and my own operating system around it.

I started by helping someone close to me put AI to use in her day-to-day, then built and tested each piece by hand before it went anywhere near a business. That is the point of working in the open: you can see the work, not just the claims.

“Instead of laying every brick, I want to design the building.”