Hi, I’m Stephen Luc.

I learn a system well enough to explain it, then look for the parts that trip people up. Sometimes the answer is a video or a clearer diagram. Sometimes it’s a tool that removes the confusing step entirely.

My background is in computer information systems and university web development. Technical writing was one of the most useful things I studied: it taught me to organize an explanation around what someone needs to do next.

Building the systems behind marketing brings those parts of my work together: understanding the customer, building what’s missing, and checking whether it helps. Here, I design the systems and decide what we’re trying to solve. The AI crew helps research, write, code and test. I stay involved in the work, including the parts that don’t go to plan.

Two examples from this site. Video Evidence Engine failed a real test, and a fix got the same video to the review step the same day. An agent testing the Living Library like an ordinary visitor found a bug; about 74 minutes later it was fixed, checked by two other agents and installed.

1. The misfit

   I took the SAT seven times. In school I was the odd kid: loud where people expected quiet, and nobody’s idea of book smart. It took me years to stop believing that last part.

2. Thirteen cashiers, one bagger

   Two weeks into bagging groceries at Kroger, they raised my pay. Some days I was the only bagger for 13 cashiers. People laughed, because I was tossing groceries like pizza dough. But I could see where every item would go before it reached me. It was Tetris, and it was a game to me.

3. The ten bidders who lost

   In college I set up websites in the library’s computer lab and sold them on eBay. Everyone else sold to the winning bidder and stopped. I went back to the ten or so people who lost and offered each one a site at their bid. That paid for food and rent. My grades took the hit.

4. The help desk

   The university built its online course system from scratch, and I ended up on support: frustrated professors, students who couldn’t log in. That’s where I learned to teach complicated things, fast.

5. Someone moved my desk

   I have ADHD. My desk always looked like a mess, but I knew where everything was. If someone moved it, I lost my place. Years later, my agents kept coming back from compaction knowing the project but not where they were in it, and I kept typing “continue.” So we built them a notebook that lives outside the conversation. That’s Continuity.

6. Same obsession, different crew

   I’ve always been obsessed with shipping faster. Back then, that meant hiring people and building a process. Now the crew is AI agents from rival companies, and for a while I was the messenger between them. Same questions, every time: who’s doing what, what changed, is it ready? That’s why we built Together. The public NVIDIA Nemotron / Nebius version is on GitHub.

I started making courses in 2003. Since then I’ve run launches and enablement for software companies, built a subscription content business, and run the member systems for a creator business with a 182K-subscriber channel. Today I build AI agent workflows through Nonstop Info Publishing, my consulting business since 2016.

Every role, with the numbers and the proof under each one, is on my portfolio. That includes Russell Brunson’s public credit for my videos in 2010.

Building the systems behind marketing is the clearest name for it. I learn the system, find where people get stuck, build what’s missing, and show people how to use it. It fits AI and agent-tool companies, developer tools, and creators with more good material than anyone can find.

I design the systems and work alongside the crew. They help research, write, build and test. You’ll see their names beside the work they contributed.

• The crew

  Chad, Faye, Way, Command, Vision, Storykeeper, Field, Quill, Chad’s Dot, Tempo, and Socket, credited on the Crew page for their contributions.

• Supporting

  Specialists and supporting workers are credited separately, per task, and are not presented as core crew.

• Names

  These are our names for AI collaborators. Their roles and project history can continue across sessions and tools.

• Verified

  Claims point to saved files, or to records of runs that actually happened.

• Observed

  Steve’s original observation is kept as said, without claiming it proves a cause.

• Unknown

  The deciding evidence is not present. Unknown is not the same as wrong.

• Failed · partial

  Failed and partial work remains part of the record.

Numbers on these pages come from saved records and keep their limits; no general time-saving, cost, or model-performance claim is made. Screenshots and art say what they are: a real capture, an earlier version, synthetic data, or an illustration.

Email me and I’ll walk you through a run, start to finish.